Side-by-side configuration is incorrect reported as permission denied

Side-by-side configuration is incorrect reported as permission denied

[Andrew DeFaria]

I use Cygwin a lot. And I kick off Windows processes a lot. Recently 
I've been having a problem with my system but from Cygwin all I see is 
"permission denied":

    Ltsdo-adefaria:cd /cygdrive/c/Program\ Files/IBM/RationalSDLC/Clearquest
    Ltsdo-adefaria:ls -l clearquest.exe
    -rwxr-xr-x+ 1 Administrators clearusers 245760 Jun  2  2011
    clearquest.exe*
    Ltsdo-adefaria:clearquest
    bash: ./clearquest: Permission denied
    Ltsdo-adefaria:

However if I use cmd the real error message comes out:

    Ltsdo-adefaria:cmd /c clearquest
    The application has failed to start because its side-by-side
    configuration is incorrect. Please see the application event log or
    use the command-line sxstrace.exe tool for more detail.

I know that this "side-by-side configuration is incorrect" is a 
configuration error on my machine and I need to fix it, but shouldn't 
Cygwin's exec(2) report the "side-by-side" error instead of the more 
erroneous "Permission denied" error?
-- 
Andrew DeFaria <http://defaria.com>
Why do you press harder on the buttons of a remote control when you know 
the batteries are dead?



--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Larry Hall (Cygwin)]

On 8/10/2012 7:31 PM, Andrew DeFaria wrote:
> I use Cygwin a lot. And I kick off Windows processes a lot. Recently I've
> been having a problem with my system but from Cygwin all I see is
> "permission denied":
>
>     Ltsdo-adefaria:cd /cygdrive/c/Program\ Files/IBM/RationalSDLC/Clearquest
>     Ltsdo-adefaria:ls -l clearquest.exe
>     -rwxr-xr-x+ 1 Administrators clearusers 245760 Jun  2  2011
>     clearquest.exe*
>     Ltsdo-adefaria:clearquest
>     bash: ./clearquest: Permission denied
>     Ltsdo-adefaria:
>
> However if I use cmd the real error message comes out:
>
>     Ltsdo-adefaria:cmd /c clearquest
>     The application has failed to start because its side-by-side
>     configuration is incorrect. Please see the application event log or
>     use the command-line sxstrace.exe tool for more detail.
>
> I know that this "side-by-side configuration is incorrect" is a
> configuration error on my machine and I need to fix it, but shouldn't
> Cygwin's exec(2) report the "side-by-side" error instead of the more
> erroneous "Permission denied" error?

Cygwin doesn't report Windows error codes.  It reports POSIX ones.  I
have no idea why there would be a POSIX error code for side-by-side
errors but if there were, then reporting that is more appropriate.

-- 
Larry

_____________________________________________________________________

A: Yes.
 > Q: Are you sure?
 >> A: Because it reverses the logical flow of conversation.
 >>> Q: Why is top posting annoying in email?

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Earnie Boyd]

On Fri, Aug 10, 2012 at 10:32 PM, Larry Hall (Cygwin) wrote:
> On 8/10/2012 7:31 PM, Andrew DeFaria wrote:
>>
>> I use Cygwin a lot. And I kick off Windows processes a lot. Recently I've
>> been having a problem with my system but from Cygwin all I see is
>> "permission denied":
>>
>>     Ltsdo-adefaria:cd /cygdrive/c/Program\
>> Files/IBM/RationalSDLC/Clearquest
>>     Ltsdo-adefaria:ls -l clearquest.exe
>>     -rwxr-xr-x+ 1 Administrators clearusers 245760 Jun  2  2011
>>     clearquest.exe*
>>     Ltsdo-adefaria:clearquest
>>     bash: ./clearquest: Permission denied
>>     Ltsdo-adefaria:
>>
>> However if I use cmd the real error message comes out:
>>
>>     Ltsdo-adefaria:cmd /c clearquest
>>     The application has failed to start because its side-by-side
>>     configuration is incorrect. Please see the application event log or
>>     use the command-line sxstrace.exe tool for more detail.
>>
>> I know that this "side-by-side configuration is incorrect" is a
>> configuration error on my machine and I need to fix it, but shouldn't
>> Cygwin's exec(2) report the "side-by-side" error instead of the more
>> erroneous "Permission denied" error?
>
>
> Cygwin doesn't report Windows error codes.  It reports POSIX ones.  I
> have no idea why there would be a POSIX error code for side-by-side
> errors but if there were, then reporting that is more appropriate.

IMO, Permission denied is correct for the error code returned from the
exec call.

Anyway http://answers.microsoft.com/en-us/windows/forum/windows_7-pictures/error-the-application-has-failed-to-start-because/df019c0d-746e-42d0-ad68-465e18e3f3ef
has the reason for the so called "side-by-side" error.

-- 
Earnie
-- https://sites.google.com/site/earnieboyd

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Andrew DeFaria]

On 08/10/2012 07:32 PM, Larry Hall (Cygwin) wrote:
> On 8/10/2012 7:31 PM, Andrew DeFaria wrote:
>> I use Cygwin a lot. And I kick off Windows processes a lot. Recently 
>> I've
>> been having a problem with my system but from Cygwin all I see is
>> "permission denied":
>>
>>     Ltsdo-adefaria:cd /cygdrive/c/Program\ 
>> Files/IBM/RationalSDLC/Clearquest
>>     Ltsdo-adefaria:ls -l clearquest.exe
>>     -rwxr-xr-x+ 1 Administrators clearusers 245760 Jun  2  2011
>>     clearquest.exe*
>>     Ltsdo-adefaria:clearquest
>>     bash: ./clearquest: Permission denied
>>     Ltsdo-adefaria:
>>
>> However if I use cmd the real error message comes out:
>>
>>     Ltsdo-adefaria:cmd /c clearquest
>>     The application has failed to start because its side-by-side
>>     configuration is incorrect. Please see the application event log or
>>     use the command-line sxstrace.exe tool for more detail.
>>
>> I know that this "side-by-side configuration is incorrect" is a
>> configuration error on my machine and I need to fix it, but shouldn't
>> Cygwin's exec(2) report the "side-by-side" error instead of the more
>> erroneous "Permission denied" error?
>
> Cygwin doesn't report Windows error codes.  It reports POSIX ones.  I
> have no idea why there would be a POSIX error code for side-by-side
> errors but if there were, then reporting that is more appropriate.
I thought that perhaps Cygwin would report back error *messages* not 
just error *codes*...
-- 
Andrew DeFaria <http://defaria.com>
The best argument against democracy is a five-minute conversation with 
the average voter.


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Christopher Faylor]

On Sat, Aug 11, 2012 at 10:32:13AM -0700, Andrew DeFaria wrote:
>On 08/10/2012 07:32 PM, Larry Hall (Cygwin) wrote:
>> On 8/10/2012 7:31 PM, Andrew DeFaria wrote:
>>> I use Cygwin a lot. And I kick off Windows processes a lot. Recently 
>>> I've
>>> been having a problem with my system but from Cygwin all I see is
>>> "permission denied":
>>>
>>>     Ltsdo-adefaria:cd /cygdrive/c/Program\ 
>>> Files/IBM/RationalSDLC/Clearquest
>>>     Ltsdo-adefaria:ls -l clearquest.exe
>>>     -rwxr-xr-x+ 1 Administrators clearusers 245760 Jun  2  2011
>>>     clearquest.exe*
>>>     Ltsdo-adefaria:clearquest
>>>     bash: ./clearquest: Permission denied
>>>     Ltsdo-adefaria:
>>>
>>> However if I use cmd the real error message comes out:
>>>
>>>     Ltsdo-adefaria:cmd /c clearquest
>>>     The application has failed to start because its side-by-side
>>>     configuration is incorrect. Please see the application event log or
>>>     use the command-line sxstrace.exe tool for more detail.
>>>
>>> I know that this "side-by-side configuration is incorrect" is a
>>> configuration error on my machine and I need to fix it, but shouldn't
>>> Cygwin's exec(2) report the "side-by-side" error instead of the more
>>> erroneous "Permission denied" error?
>>
>> Cygwin doesn't report Windows error codes.  It reports POSIX ones.  I
>> have no idea why there would be a POSIX error code for side-by-side
>> errors but if there were, then reporting that is more appropriate.
>I thought that perhaps Cygwin would report back error *messages* not 
>just error *codes*...

Cygwin emulates Linux.  "Permission denied" is an error message associated
with a specific errno.  Neither Cygwin nor Linux know anything about a
"side-by-side" configuration problem.

cgf

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Cliff Hones]

On 11/08/2012 20:22, Christopher Faylor wrote:
> On Sat, Aug 11, 2012 at 10:32:13AM -0700, Andrew DeFaria wrote:
>> On 08/10/2012 07:32 PM, Larry Hall (Cygwin) wrote:
>>> On 8/10/2012 7:31 PM, Andrew DeFaria wrote:
>>>> I use Cygwin a lot. And I kick off Windows processes a lot. Recently 
>>>> I've
>>>> been having a problem with my system but from Cygwin all I see is
>>>> "permission denied":
>>>>
>>>>     Ltsdo-adefaria:cd /cygdrive/c/Program\ 
>>>> Files/IBM/RationalSDLC/Clearquest
>>>>     Ltsdo-adefaria:ls -l clearquest.exe
>>>>     -rwxr-xr-x+ 1 Administrators clearusers 245760 Jun  2  2011
>>>>     clearquest.exe*
>>>>     Ltsdo-adefaria:clearquest
>>>>     bash: ./clearquest: Permission denied
>>>>     Ltsdo-adefaria:
>>>>
>>>> However if I use cmd the real error message comes out:
>>>>
>>>>     Ltsdo-adefaria:cmd /c clearquest
>>>>     The application has failed to start because its side-by-side
>>>>     configuration is incorrect. Please see the application event log or
>>>>     use the command-line sxstrace.exe tool for more detail.
>>>>
>>>> I know that this "side-by-side configuration is incorrect" is a
>>>> configuration error on my machine and I need to fix it, but shouldn't
>>>> Cygwin's exec(2) report the "side-by-side" error instead of the more
>>>> erroneous "Permission denied" error?
>>>
>>> Cygwin doesn't report Windows error codes.  It reports POSIX ones.  I
>>> have no idea why there would be a POSIX error code for side-by-side
>>> errors but if there were, then reporting that is more appropriate.
>> I thought that perhaps Cygwin would report back error *messages* not 
>> just error *codes*...
> 
> Cygwin emulates Linux.  "Permission denied" is an error message associated
> with a specific errno.  Neither Cygwin nor Linux know anything about a
> "side-by-side" configuration problem.

I imagine there are many Windows errors which Cygwin has to interpret internally
and present as POSIX errors to the user.  It looks like this Windows7 error
may mean that some Windows runtime components are missing - in which case
wouldn't ELIBACC or ELIBEXEC be more appropriate?  "Permission denied" suggests
that changing the access permissions, or running as a user with greater
privileges would solve the problem.

-- Cliff


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Andrew DeFaria]

On 08/11/2012 12:22 PM, Christopher Faylor wrote:
> On Sat, Aug 11, 2012 at 10:32:13AM -0700, Andrew DeFaria wrote:
>> On 08/10/2012 07:32 PM, Larry Hall (Cygwin) wrote:
>>> On 8/10/2012 7:31 PM, Andrew DeFaria wrote:
>>>> I use Cygwin a lot. And I kick off Windows processes a lot. Recently
>>>> I've
>>>> been having a problem with my system but from Cygwin all I see is
>>>> "permission denied":
>>>>
>>>>      Ltsdo-adefaria:cd /cygdrive/c/Program\
>>>> Files/IBM/RationalSDLC/Clearquest
>>>>      Ltsdo-adefaria:ls -l clearquest.exe
>>>>      -rwxr-xr-x+ 1 Administrators clearusers 245760 Jun  2  2011
>>>>      clearquest.exe*
>>>>      Ltsdo-adefaria:clearquest
>>>>      bash: ./clearquest: Permission denied
>>>>      Ltsdo-adefaria:
>>>>
>>>> However if I use cmd the real error message comes out:
>>>>
>>>>      Ltsdo-adefaria:cmd /c clearquest
>>>>      The application has failed to start because its side-by-side
>>>>      configuration is incorrect. Please see the application event log or
>>>>      use the command-line sxstrace.exe tool for more detail.
>>>>
>>>> I know that this "side-by-side configuration is incorrect" is a
>>>> configuration error on my machine and I need to fix it, but shouldn't
>>>> Cygwin's exec(2) report the "side-by-side" error instead of the more
>>>> erroneous "Permission denied" error?
>>> Cygwin doesn't report Windows error codes.  It reports POSIX ones.  I
>>> have no idea why there would be a POSIX error code for side-by-side
>>> errors but if there were, then reporting that is more appropriate.
>> I thought that perhaps Cygwin would report back error *messages* not
>> just error *codes*...
> Cygwin emulates Linux.  "Permission denied" is an error message associated
> with a specific errno.  Neither Cygwin nor Linux know anything about a
> "side-by-side" configuration problem.
Which is really all the more reason to output the Windows error message 
along with perhaps the status code. That is if the goal is 
understandability. If the goal is obfuscation then I'd say your well on 
your way to hitting your mark. ;-)
-- 
Andrew DeFaria <http://defaria.com>
Your child may be an honor student but you're still an idiot.


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Pawel Jasinski]

Hi,

>
> Cygwin doesn't report Windows error codes.  It reports POSIX ones.  I
> have no idea why there would be a POSIX error code for side-by-side
> errors but if there were, then reporting that is more appropriate.
>

out of curiosity I have poked into errno.h (as well as man page) on
linux and linux is easier on its users.
It uses values defined by POSIX as well as values which make sense in
context of Linux.
Would you consider relaxing your policy in respect of error codes and
POSIX where would it be a benefit for the user?

In this particular case, all situation where errno (or a message) is
simply reported back to the user would be fine.
Only code which does errno specific attempts to recover or be clever
about it would be affected.
At the moment I my internet access is weak, but once I am at home this
would be a grep through cygwin packages to find out such a use.

Cheers,
Pawel

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Christopher Faylor]

On Sun, Aug 12, 2012 at 11:07:00AM +0200, Pawel Jasinski wrote:
>Hi,
>
>>
>> Cygwin doesn't report Windows error codes.  It reports POSIX ones.  I
>> have no idea why there would be a POSIX error code for side-by-side
>> errors but if there were, then reporting that is more appropriate.
>>
>
>out of curiosity I have poked into errno.h (as well as man page) on
>linux and linux is easier on its users.
>It uses values defined by POSIX as well as values which make sense in
>context of Linux.
>Would you consider relaxing your policy in respect of error codes and
>POSIX where would it be a benefit for the user?
>
>In this particular case, all situation where errno (or a message) is
>simply reported back to the user would be fine.
>Only code which does errno specific attempts to recover or be clever
>about it would be affected.

Yes, and, what do we do with these "clever" apps?  Tell them that they
are out of luck because we've just pulled the rug out from under them?

>At the moment I my internet access is weak, but once I am at home this
>would be a grep through cygwin packages to find out such a use.

Windows has 15999 error codes.  Linux has ~133.  Some UNIX apps which
recognize errnos rely on being able to index into the "sys_errlist"
array.  This would mean that we would need to set up a sys_errlist
array with all of these error codes, and since we already map a lot
of them, come up with some way to deal with that.

If you can find a nice Linux errno which maps from ERROR_SXS_CANT_GEN_ACTCTX
to something other than EACCES I'd be happy to change Cygwin.  Otherwise,
no, I'm not going to worry about this issue.

cgf

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Pawel Jasinski]

hi,

On Sun, Aug 12, 2012 at 7:06 PM, Christopher Faylor wrote:
> On Sun, Aug 12, 2012 at 11:07:00AM +0200, Pawel Jasinski wrote:
>>Hi,
>>
>>>
>>> Cygwin doesn't report Windows error codes.  It reports POSIX ones.  I
>>> have no idea why there would be a POSIX error code for side-by-side
>>> errors but if there were, then reporting that is more appropriate.
>>>
>>
>>out of curiosity I have poked into errno.h (as well as man page) on
>>linux and linux is easier on its users.
>>It uses values defined by POSIX as well as values which make sense in
>>context of Linux.
>>Would you consider relaxing your policy in respect of error codes and
>>POSIX where would it be a benefit for the user?
>>
>>In this particular case, all situation where errno (or a message) is
>>simply reported back to the user would be fine.
>>Only code which does errno specific attempts to recover or be clever
>>about it would be affected.
>
> Yes, and, what do we do with these "clever" apps?  Tell them that they
> are out of luck because we've just pulled the rug out from under them?


For the clever apps, I wonder what the app can do about no permissions
anyway other than report it.
And since the problem reported now as permission problem has nothing
to do with it, the corrective actions are adding to the confusion.

Most of the one I know report errno and eventually error message and
say bye bye. It is up to you to sudo or "run as".
I am still not at home, so no stats from the packages.

>
>>At the moment I my internet access is weak, but once I am at home this
>>would be a grep through cygwin packages to find out such a use.
>
> Windows has 15999 error codes.  Linux has ~133.  Some UNIX apps which
> recognize errnos rely on being able to index into the "sys_errlist"
> array.  This would mean that we would need to set up a sys_errlist
> array with all of these error codes, and since we already map a lot
> of them, come up with some way to deal with that.

wow, the number of error codes is impressive.
But we are talking about one, not 15999 minus whatever is already mapped.
How about taking care of the things which annoys your users.
This potentially would have a very positive side effect, decreased
traffic on this mailing list.

>
> If you can find a nice Linux errno which maps from ERROR_SXS_CANT_GEN_ACTCTX
> to something other than EACCES I'd be happy to change Cygwin.

Sorry I was not clear. The point was not to use linux error code. what
I was trying to say is linux does not follow POSIX only error codes
policy.
Such a policy is impractical. Your point was POSIX only, wasn't it?
You can add cygwin specific error codes.


> Otherwise,
> no, I'm not going to worry about this issue.

Could you please qualify your no.
No, as long as you have to do it yourself.
No, no patches will be considered.
No, even if the request comes from red hat support channel (I consider
buying one, I wonder what can I get for the money).


Cheers,
Pawel

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Earnie Boyd]

On Sun, Aug 12, 2012 at 1:56 PM, Pawel Jasinski wrote:
>>
>> If you can find a nice Linux errno which maps from ERROR_SXS_CANT_GEN_ACTCTX
>> to something other than EACCES I'd be happy to change Cygwin.
>
> Sorry I was not clear. The point was not to use linux error code. what
> I was trying to say is linux does not follow POSIX only error codes
> policy.
> Such a policy is impractical. Your point was POSIX only, wasn't it?
> You can add cygwin specific error codes.

Cygwin doesn't have the access required to execute the process because
of some reason.  It is up to you to examine the reason.  Why should we
be bothered with anything other than EACCESS?  When you reported the
issue the first action you took was to determine why you couldn't
execute a Windows program.  That was the correct action, hammering
this list with your wishes isn't gaining you sympathy from those that
use and support it.

-- 
Earnie
-- https://sites.google.com/site/earnieboyd

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Christopher Faylor]

On Sun, Aug 12, 2012 at 04:35:23PM -0400, Earnie Boyd wrote:
>On Sun, Aug 12, 2012 at 1:56 PM, Pawel Jasinski wrote:
>>>If you can find a nice Linux errno which maps from
>>>ERROR_SXS_CANT_GEN_ACTCTX to something other than EACCES I'd be happy
>>>to change Cygwin.
>>
>>Sorry I was not clear.  The point was not to use linux error code.
>>what I was trying to say is linux does not follow POSIX only error
>>codes policy.  Such a policy is impractical.  Your point was POSIX
>>only, wasn't it?  You can add cygwin specific error codes.
>
>Cygwin doesn't have the access required to execute the process because
>of some reason.  It is up to you to examine the reason.  Why should we
>be bothered with anything other than EACCESS?  When you reported the
>issue the first action you took was to determine why you couldn't
>execute a Windows program.  That was the correct action, hammering this
>list with your wishes isn't gaining you sympathy from those that use
>and support it.

Actually, it was Andrew Defaria who wanted it but, yeah, I think I've
personally devoted more time to this than the issue warrants.

I've asked Corinna privately if she has any ideas or concerns but I
would be surprised if this was something that she thought important
to address.

If someone has a patch they want to propose they should send it to
cygwin-patches, bearing in mind that we don't want to break any
existing applications.

cgf

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Pawel Jasinski]

hi,

Thanks Christopher and Earnie for clearing things up.
I am trying to promote cygwin as a solution to make working with
windows better than windows.
Unfortunately I was hit by permission problem (exactly as described by
Andrew) in front of other people.
Let me say this, it did not help to promote cygwin.
I hope this can be seen as a bit of justification for my strong argument.

Cheers,
Pawel

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Andrew DeFaria]

On 08/12/2012 01:35 PM, Earnie Boyd wrote:
> On Sun, Aug 12, 2012 at 1:56 PM, Pawel Jasinski wrote:
>>> If you can find a nice Linux errno which maps from ERROR_SXS_CANT_GEN_ACTCTX
>>> to something other than EACCES I'd be happy to change Cygwin.
>> Sorry I was not clear. The point was not to use linux error code. what
>> I was trying to say is linux does not follow POSIX only error codes
>> policy.
>> Such a policy is impractical. Your point was POSIX only, wasn't it?
>> You can add cygwin specific error codes.
Although I believe I understand some of the issues of POSIX or 
POSIX-like compliance in reporting errnos, etc. I still wish to address 
the following.
> Cygwin doesn't have the access required to execute the process because
> of some reason.
Exactly.
> It is up to you to examine the reason.
Yes and no, IMHO. You (meaning Cygwin) have immensely more information 
about what's going on than I (the user) do! You are the best person in 
the whole world at this time to report what went wrong so that I can fix 
the problem and return to profitability. And you're gonna hoard that info???
> Why should we be bothered with anything other than EACCESS?
Like I said, because you have much more information about what's going 
on, specifically what you were doing and where in the series of calls 
and instructions that lead you to an error. True some other sub-system 
may have errored out and perhaps didn't even give you much to go on, but 
in general you'd have more information about what went wrong than the user.
> When you reported the issue the first action you took was to determine why you couldn't
> execute a Windows program. That was the correct action,
As the actual person who reported the issue I can assure you that my 
first action was most definitely not to determine why I couldn't execute 
a Windows program. The reason seemed clear - or at least was clearly 
reported to me although the reason was actually erroneous. I truly 
thought it was a permission problem. I checked the permission bits and 
wondered what was going on because according to Cygwin I *HAD* 
permissions to execute the file. I thought perhaps it was on a share or 
maybe the Windows permissions matrix was weirdly configured, etc. When 
none of that panned out I got the bright idea to try it in cmd, which 
gave me the better error message. Then I scratched my head and wondered 
by Cygwin didn't simply report that error message (that I assume 
CreateProcess passed back to it - however I don't write Windows code) 
and simply and dutifully report back the text of the error message the 
OS gave it. Many other users would not have thought to try that.

Now I know that there are POSIX error codes and people tend to check 
error numbers and you want to put the closest match of a POSIX errno to 
the Windows error message. But why couldn't you just print to stderr the 
error message that Windows returned perhaps in addition to setting 
syserr to "Permission denied"?
> hammering this list with your wishes isn't gaining you sympathy from those that
> use and support it.
I don't know. He has my sympathy. My pity too... :-)
-- 
Andrew DeFaria <http://defaria.com>
The other day I was playing poker with Tarot cards. I got a full house 
and four people died.


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Herbert Stocker]

Hi,

Imho, EACCESS is indeed a bit misleading because it suggests permission
problems. Better would be to have an EFAIL as a generic error. Actually i
was missing an EFAIL several times when my programs needed to return
an error code that did not match well with what i found in errno.h .

On Sun, Aug 12, 2012 at 11:07:00AM +0200, Pawel Jasinski wrote:

> If you can find a nice Linux errno which maps from ERROR_SXS_CANT_GEN_ACTCTX
> to something other than EACCES I'd be happy to change Cygwin.

I had a quick poke into sys/errno.h and there i found 5 error codes
beginning with ELIB. One of those should suffice. My fafourite is this:

    #define ELIBBAD 84.     /* Accessing a corrupted shared lib */

Because side-by-side problems may mean that the supporting DLL is
acutally there and can be read (also for execute), but the accompanying
XML file describes it incorrectly (e.g. wrong version number), the DLL
is not signed correctly, is not placed in the subdirectory whose name
is mandated by Windows, etc.

> Otherwise, no, I'm not going to worry about this issue.

There is no need for the 'no', i'd suggest ELIBBAD.

(And to bring back to memory, i'd also suggest to add EFAIL be added)


Herbert

P.S.: thanks for the hint, Pawel.



--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Corinna Vinschen]

On Aug 12 18:49, Andrew DeFaria wrote:
> On 08/12/2012 01:35 PM, Earnie Boyd wrote:
> >On Sun, Aug 12, 2012 at 1:56 PM, Pawel Jasinski wrote:
> >>>If you can find a nice Linux errno which maps from ERROR_SXS_CANT_GEN_ACTCTX
> >>>to something other than EACCES I'd be happy to change Cygwin.
> [...]
> Now I know that there are POSIX error codes and people tend to check
> error numbers and you want to put the closest match of a POSIX errno
> to the Windows error message. But why couldn't you just print to
> stderr the error message that Windows returned perhaps in addition
> to setting syserr to "Permission denied"?

Do you really want that?

There's a difference between cmd and Cygwin.  Cmd is a shell, Cygwin is
just the underlying shared lib providing a generic API.

If an error occurs, it's the shell's responsibility to print an error
message in the first place.  All messages printed by Cygwin are not
controllable by the calling application.  Therefore we usually only
print messages from the DLL if something very serious happens from the
DLLs perspective.  Some arbitrary Windows error code returned from
CreateProcess is usually not something actually serious.  There was
just "some" reason that an application couldn't be started.

Also, where do you draw the border?  Which windows error code is serious
enough to justify a (pretty intrusive!) error message from the underlying
library and which isn't?

As cgf pointed out, Windows has zillions of error codes.  We wouldn't
want to generate the same number of POSIX-like error codes.  It wouldn't
make a lot of sense since POSIX applications only test for a limited,
expected number of error codes, and it might break things.

Having said that, EACCES is not bad as far as error codes go in this
case, but I like Cliff Hones' idea to choose another existing POSIX
error like ELIBACC, "Can not access a needed shared library", which
might make more sense in this scenario.  I'd also like ENOPKG, "Package
not installed".


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader          cygwin AT cygwin DOT com
Red Hat

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Corinna Vinschen]

On Aug 13 10:24, Herbert Stocker wrote:
> There is no need for the 'no', i'd suggest ELIBBAD.

Not bad, either.  So we have ELIBACC, ELIBBADD, and ENOPKG as 
suggestions.

> (And to bring back to memory, i'd also suggest to add EFAIL be added)

Here's a clear "no".  EFAIL is not a useful error message.  It's not
even slightly wrong, like EACCES in this case might be, it's entirely
lacking information.  What's the accompanying error message?  "Something
failed"?  "An error occured, but I have no idea what"?

This reminds me of the old awk, which apparently only knew two error
messages:

  awk: syntax error near line x
  awk: bailing out near line x

In which scenario would such an EFAIL be really useful, which wouldn't
be better covered by an error code which contains at least *some*
information about the cause?


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader          cygwin AT cygwin DOT com
Red Hat

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Herbert Stocker]

On 13.08.2012 10:47, Corinna Vinschen wrote:
> On Aug 13 10:24, Herbert Stocker wrote:
>> There is no need for the 'no', i'd suggest ELIBBAD.
> Not bad, either.  So we have ELIBACC, ELIBBADD, and ENOPKG as
> suggestions.
I'd vote against ENOPKG. That means: /* Package not installed */
A side-by-side error does not primarily mean that something is not 
installed.

>> (And to bring back to memory, i'd also suggest to add EFAIL be added)
> Here's a clear "no".  EFAIL is not a useful error message.  It's not
> even slightly wrong, like EACCES in this case might be, it's entirely
> lacking information.
My suggestion was not to propose EFAIL as an error message for a 
side-by-side error.
Not at all.
i should have made an extra toipic, actually, and if i would consider it 
important enough, i
would make one now.

What i meant was that sometimes i'm writing a program, and when 
searching for an
appropriate error code in errno.h, i don't find one. Then i would prefer 
to resort to some
genereric EFAIL.
Imagine a routine decodes an mp3 stream and fails because the mp3 data 
is corrupt.
What error code would match this? The best i found was:
    #define EBADMSG 77.     /* Trying to read unreadable message */

But i guess the solution would be to stay away from these more 
OS-centric error codes.

Again, i do not have a strong desire on discussing EFAIL, so we can stop 
here.

Herbert



--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Earnie Boyd]

On Mon, Aug 13, 2012 at 4:24 AM, Herbert Stocker wrote:
> Hi,
>
> Imho, EACCESS is indeed a bit misleading because it suggests permission
> problems. Better would be to have an EFAIL as a generic error. Actually i
> was missing an EFAIL several times when my programs needed to return
> an error code that did not match well with what i found in errno.h .

You may think it is misleading but
http://pubs.opengroup.org/onlinepubs/009604499/functions/exec.html
states that EACCESS is the correct value.

>
>
> On Sun, Aug 12, 2012 at 11:07:00AM +0200, Pawel Jasinski wrote:
>
>> If you can find a nice Linux errno which maps from
>> ERROR_SXS_CANT_GEN_ACTCTX
>> to something other than EACCES I'd be happy to change Cygwin.
>
>
> I had a quick poke into sys/errno.h and there i found 5 error codes
> beginning with ELIB. One of those should suffice. My fafourite is this:
>

Not valid for exec() to return.

>    #define ELIBBAD 84.     /* Accessing a corrupted shared lib */
>
> Because side-by-side problems may mean that the supporting DLL is
> acutally there and can be read (also for execute), but the accompanying
> XML file describes it incorrectly (e.g. wrong version number), the DLL
> is not signed correctly, is not placed in the subdirectory whose name
> is mandated by Windows, etc.
>
>> Otherwise, no, I'm not going to worry about this issue.
>
>
> There is no need for the 'no', i'd suggest ELIBBAD.

Again, not an error message that exec() should return.

-- 
Earnie
-- https://sites.google.com/site/earnieboyd

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Corinna Vinschen]

On Aug 13 08:51, Earnie Boyd wrote:
> On Mon, Aug 13, 2012 at 4:24 AM, Herbert Stocker wrote:
> > There is no need for the 'no', i'd suggest ELIBBAD.
> 
> Again, not an error message that exec() should return.

Wrong.  See the Linux man page for execve:

http://linux.die.net/man/2/execve

  ELIBBAD
      An ELF interpreter was not in a recognized format. 


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader          cygwin AT cygwin DOT com
Red Hat

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Earnie Boyd]

On Mon, Aug 13, 2012 at 9:11 AM, Corinna Vinschen> wrote:
> On Aug 13 08:51, Earnie Boyd wrote:
>> On Mon, Aug 13, 2012 at 4:24 AM, Herbert Stocker wrote:
>> > There is no need for the 'no', i'd suggest ELIBBAD.
>>
>> Again, not an error message that exec() should return.
>
> Wrong.  See the Linux man page for execve:
>
> http://linux.die.net/man/2/execve
>
>   ELIBBAD
>       An ELF interpreter was not in a recognized format.

That page also says

EACCES
   Execute permission is denied for the file or a script or ELF interpreter.

I don't see this SxS error an unrecognized format.  The users system
is rather incomplete causing the execution to fail.  The MS solution
is to install the redistributable DLL files.

-- 
Earnie
-- https://sites.google.com/site/earnieboyd

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Cliff Hones]

On 13/08/2012 13:51, Earnie Boyd wrote:
> On Mon, Aug 13, 2012 at 4:24 AM, Herbert Stocker wrote:
>> Hi,
>>
>> Imho, EACCESS is indeed a bit misleading because it suggests permission
>> problems. Better would be to have an EFAIL as a generic error. Actually i
>> was missing an EFAIL several times when my programs needed to return
>> an error code that did not match well with what i found in errno.h .
> 
> You may think it is misleading but
> http://pubs.opengroup.org/onlinepubs/009604499/functions/exec.html
> states that EACCESS is the correct value.

Well, for a start that's an POSIX V1.  Here's a link to V2 exec:

   http://pubs.opengroup.org/onlinepubs/9699919799/functions/exec.html

I don't see in the description for EACCES [Note - one S, not two]
that it should be returned when there is a missing runtime component.

Also, see this general page on errors.  The errors documented for a
particular function aren't intended to be exhaustive.  An implementation
can return others as long as it is consistent.

   http://pubs.opengroup.org/onlinepubs/9699919799/functions/V2_chap02.html#tag_15_03

In any case Cygwin's primary aim is to provide a Linux-like environment,
not pure POSIX.  Linux exec/execve manpages list many more error codes.

-- Cliff


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Corinna Vinschen]

On Aug 13 09:21, Earnie Boyd wrote:
> On Mon, Aug 13, 2012 at 9:11 AM, Corinna Vinschen> wrote:
> > On Aug 13 08:51, Earnie Boyd wrote:
> >> On Mon, Aug 13, 2012 at 4:24 AM, Herbert Stocker wrote:
> >> > There is no need for the 'no', i'd suggest ELIBBAD.
> >>
> >> Again, not an error message that exec() should return.
> >
> > Wrong.  See the Linux man page for execve:
> >
> > http://linux.die.net/man/2/execve
> >
> >   ELIBBAD
> >       An ELF interpreter was not in a recognized format.
> 
> That page also says
> 
> EACCES
>    Execute permission is denied for the file or a script or ELF interpreter.
> 
> I don't see this SxS error an unrecognized format.  The users system
> is rather incomplete causing the execution to fail.  The MS solution
> is to install the redistributable DLL files.

What I was trying to say is, Linux allows execve to return POSIX error
codes which are not covered by the POSIX specs of execve.  EACCES is a
perfectly valid error code here, but we can also use another one for
border cases, like the suggested one.

Please note, I'm neither strongly for nor against changing the error
code in the case discussed in this thread.  But if it can help users
in a border case...


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader          cygwin AT cygwin DOT com
Red Hat

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Andrew DeFaria]

On 08/13/2012 01:27 AM, Corinna Vinschen wrote:
> On Aug 12 18:49, Andrew DeFaria wrote:
>> On 08/12/2012 01:35 PM, Earnie Boyd wrote:
>>> On Sun, Aug 12, 2012 at 1:56 PM, Pawel Jasinski wrote:
>>>>> If you can find a nice Linux errno which maps from ERROR_SXS_CANT_GEN_ACTCTX
>>>>> to something other than EACCES I'd be happy to change Cygwin.
>> [...]
>> Now I know that there are POSIX error codes and people tend to check
>> error numbers and you want to put the closest match of a POSIX errno
>> to the Windows error message. But why couldn't you just print to
>> stderr the error message that Windows returned perhaps in addition
>> to setting syserr to "Permission denied"?
> Do you really want that?
I think it would be helpful.
> There's a difference between cmd and Cygwin.  Cmd is a shell, Cygwin is
> just the underlying shared lib providing a generic API.
OK so bash...
> If an error occurs, it's the shell's responsibility to print an error
> message in the first place.  All messages printed by Cygwin are not
> controllable by the calling application.  Therefore we usually only
> print messages from the DLL if something very serious happens from the
> DLLs perspective.  Some arbitrary Windows error code returned from
> CreateProcess is usually not something actually serious.  There was
> just "some" reason that an application couldn't be started.
IMHO "some" reason that the user should be alerted about. How is it 
helpful to the end user to suppress the error message?
> Also, where do you draw the border?  Which windows error code is serious
> enough to justify a (pretty intrusive!) error message from the underlying
> library and which isn't?
I would draw the border at "if there's an error message".
> As cgf pointed out, Windows has zillions of error codes.  We wouldn't
> want to generate the same number of POSIX-like error codes.  It wouldn't
> make a lot of sense since POSIX applications only test for a limited,
> expected number of error codes, and it might break things.
I was talking error *messages* not error *codes*.
> Having said that, EACCES is not bad as far as error codes go in this
> case, but I like Cliff Hones' idea to choose another existing POSIX
> error like ELIBACC, "Can not access a needed shared library", which
> might make more sense in this scenario.  I'd also like ENOPKG, "Package
> not installed".
Again, I was talking error messages not error codes.
-- 
Andrew DeFaria <http://defaria.com>
Why is there only one Monopolies commission?


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Corinna Vinschen]

On Aug 13 07:32, Andrew DeFaria wrote:
> On 08/13/2012 01:27 AM, Corinna Vinschen wrote:
> >On Aug 12 18:49, Andrew DeFaria wrote:
> >>On 08/12/2012 01:35 PM, Earnie Boyd wrote:
> >>>On Sun, Aug 12, 2012 at 1:56 PM, Pawel Jasinski wrote:
> >>>>>If you can find a nice Linux errno which maps from ERROR_SXS_CANT_GEN_ACTCTX
> >>>>>to something other than EACCES I'd be happy to change Cygwin.
> >>[...]
> >>Now I know that there are POSIX error codes and people tend to check
> >>error numbers and you want to put the closest match of a POSIX errno
> >>to the Windows error message. But why couldn't you just print to
> >>stderr the error message that Windows returned perhaps in addition
> >>to setting syserr to "Permission denied"?
> >Do you really want that?
> I think it would be helpful.
> >There's a difference between cmd and Cygwin.  Cmd is a shell, Cygwin is
> >just the underlying shared lib providing a generic API.
> OK so bash...

Ok so bash what?

> >If an error occurs, it's the shell's responsibility to print an error
> >message in the first place.  All messages printed by Cygwin are not
> >controllable by the calling application.  Therefore we usually only
> >print messages from the DLL if something very serious happens from the
> >DLLs perspective.  Some arbitrary Windows error code returned from
> >CreateProcess is usually not something actually serious.  There was
> >just "some" reason that an application couldn't be started.
> IMHO "some" reason that the user should be alerted about. How is it
> helpful to the end user to suppress the error message?

Huh?  Somehow you swapping cause and effect.  There is no "suppressed"
error message.  Generating an error message is the task of the shell in
the first place.

It's not that the OS generates an error message and cmd lets it slip
through while Cygwin (or bash) "suppress" it.  It's the CreateProcess
call which generates an error code ERROR_SXS_CANT_GEN_ACTCTX and cmd
printing the connected error message, just like bash gets an error code
EACCES and prints the connected error message "Permission denied".

> >Also, where do you draw the border?  Which windows error code is serious
> >enough to justify a (pretty intrusive!) error message from the underlying
> >library and which isn't?
> I would draw the border at "if there's an error message".

Again, cause and effect turned upside down.

> >As cgf pointed out, Windows has zillions of error codes.  We wouldn't
> >want to generate the same number of POSIX-like error codes.  It wouldn't
> >make a lot of sense since POSIX applications only test for a limited,
> >expected number of error codes, and it might break things.
> I was talking error *messages* not error *codes*.

Again, cause and effect turned upside down.

> >Having said that, EACCES is not bad as far as error codes go in this
> >case, but I like Cliff Hones' idea to choose another existing POSIX
> >error like ELIBACC, "Can not access a needed shared library", which
> >might make more sense in this scenario.  I'd also like ENOPKG, "Package
> >not installed".
> Again, I was talking error messages not error codes.

Again, cause and effect turned upside down.


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader          cygwin AT cygwin DOT com
Red Hat

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Andrew DeFaria]

On 08/13/2012 07:42 AM, Corinna Vinschen wrote:
>>> There's a difference between cmd and Cygwin.  Cmd is a shell, Cygwin is
>>> just the underlying shared lib providing a generic API.
>> OK so bash...
> Ok so bash what?
You were saying cmd is a shell and Cygwin is a shared lib. So then 
perhaps the responsibility should fall to bash - a shell...
>>> If an error occurs, it's the shell's responsibility to print an error
>>> message in the first place.  All messages printed by Cygwin are not
>>> controllable by the calling application.  Therefore we usually only
>>> print messages from the DLL if something very serious happens from the
>>> DLLs perspective.  Some arbitrary Windows error code returned from
>>> CreateProcess is usually not something actually serious.  There was
>>> just "some" reason that an application couldn't be started.
>> IMHO "some" reason that the user should be alerted about. How is it
>> helpful to the end user to suppress the error message?
> Huh?  Somehow you swapping cause and effect.  There is no "suppressed"
> error message.  Generating an error message is the task of the shell in
> the first place.
There was an error message that cmd showed that bash did not. To me 
that's suppression.
> It's not that the OS generates an error message and cmd lets it slip
> through while Cygwin (or bash) "suppress" it.  It's the CreateProcess
> call which generates an error code ERROR_SXS_CANT_GEN_ACTCTX and cmd
> printing the connected error message, just like bash gets an error code
> EACCES and prints the connected error message "Permission denied".
Plumbing and mechanics aside, I'm just saying the user should be told 
the underlying problem. If ERROR_SXS_CANT_GEN_ACTCTX is the error code 
could ya at least print that as a string? It would give the user a 
fighting change and finding a solution...
>>> Also, where do you draw the border?  Which windows error code is serious
>>> enough to justify a (pretty intrusive!) error message from the underlying
>>> library and which isn't?
>> I would draw the border at "if there's an error message".
> Again, cause and effect turned upside down.
Not really. You seem stuck in thinking in only error codes - I'm 
thinking in solutions, regardless of the mechanics to get there. IOW the 
goal is to inform the user not only that an error occurred but what that 
errors was so that the user can fix it. How you accomplish that is not 
important (to me - speaking as an end user).

(Note: I understand the underlying issues somewhat - I'm just giving you 
a user's perspective).
>>> As cgf pointed out, Windows has zillions of error codes.  We wouldn't
>>> want to generate the same number of POSIX-like error codes.  It wouldn't
>>> make a lot of sense since POSIX applications only test for a limited,
>>> expected number of error codes, and it might break things.
>> I was talking error *messages* not error *codes*.
> Again, cause and effect turned upside down.
I don't think so.
-- 
Andrew DeFaria <http://defaria.com>
Why do you park in a driveway and drive on a parkway?


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Christopher Faylor]

On Mon, Aug 13, 2012 at 10:24:53AM +0200, Herbert Stocker wrote:
>Hi,
>
>Imho, EACCESS is indeed a bit misleading because it suggests permission
>problems. Better would be to have an EFAIL as a generic error. Actually i
>was missing an EFAIL several times when my programs needed to return
>an error code that did not match well with what i found in errno.h .
>
>On Sun, Aug 12, 2012 at 11:07:00AM +0200, Pawel Jasinski wrote:
>
>> If you can find a nice Linux errno which maps from ERROR_SXS_CANT_GEN_ACTCTX
>> to something other than EACCES I'd be happy to change Cygwin.
>
>I had a quick poke into sys/errno.h and there i found 5 error codes
>beginning with ELIB. One of those should suffice. My fafourite is this:
>
>    #define ELIBBAD 84.     /* Accessing a corrupted shared lib */
>
>Because side-by-side problems may mean that the supporting DLL is
>acutally there and can be read (also for execute), but the accompanying
>XML file describes it incorrectly (e.g. wrong version number), the DLL
>is not signed correctly, is not placed in the subdirectory whose name
>is mandated by Windows, etc.
>
>> Otherwise, no, I'm not going to worry about this issue.
>
>There is no need for the 'no', i'd suggest ELIBBAD.

"No need for the 'no'"?  Huh?  Huh?  I ASKED for a suggestion.  You
provided one.  You understand what the word "Otherwise" means right?

cgf

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Corinna Vinschen]

On Aug 13 07:59, Andrew DeFaria wrote:
> On 08/13/2012 07:42 AM, Corinna Vinschen wrote:
> >>>There's a difference between cmd and Cygwin.  Cmd is a shell, Cygwin is
> >>>just the underlying shared lib providing a generic API.
> >>OK so bash...
> >Ok so bash what?
> You were saying cmd is a shell and Cygwin is a shared lib. So then
> perhaps the responsibility should fall to bash - a shell...
> >>>If an error occurs, it's the shell's responsibility to print an error
> >>>message in the first place.  All messages printed by Cygwin are not
> >>>controllable by the calling application.  Therefore we usually only
> >>>print messages from the DLL if something very serious happens from the
> >>>DLLs perspective.  Some arbitrary Windows error code returned from
> >>>CreateProcess is usually not something actually serious.  There was
> >>>just "some" reason that an application couldn't be started.
> >>IMHO "some" reason that the user should be alerted about. How is it
> >>helpful to the end user to suppress the error message?
> >Huh?  Somehow you swapping cause and effect.  There is no "suppressed"
> >error message.  Generating an error message is the task of the shell in
> >the first place.
> There was an error message that cmd showed that bash did not. To me
> that's suppression.

This is nonsense and you know it.

> >It's not that the OS generates an error message and cmd lets it slip
> >through while Cygwin (or bash) "suppress" it.  It's the CreateProcess
> >call which generates an error code ERROR_SXS_CANT_GEN_ACTCTX and cmd
> >printing the connected error message, just like bash gets an error code
> >EACCES and prints the connected error message "Permission denied".
> Plumbing and mechanics aside, I'm just saying the user should be
> told the underlying problem. If ERROR_SXS_CANT_GEN_ACTCTX is the
> error code could ya at least print that as a string? It would give
> the user a fighting change and finding a solution...

Plumbing and mechanics is the way to solve the problem.  If you don't
want to leave the user-only perspective, we will have a hard time to
find a solution.

> >>I was talking error *messages* not error *codes*.
> >Again, cause and effect turned upside down.
> I don't think so.

We can circle all day around this point.  If you don't want to
discuss this problem seriously, I'll drop off from this thread.


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader          cygwin AT cygwin DOT com
Red Hat

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Christopher Faylor]

On Mon, Aug 13, 2012 at 07:32:11AM -0700, Andrew DeFaria wrote:
>On 08/13/2012 01:27 AM, Corinna Vinschen wrote:
>> On Aug 12 18:49, Andrew DeFaria wrote:
>>> On 08/12/2012 01:35 PM, Earnie Boyd wrote:
>>>> On Sun, Aug 12, 2012 at 1:56 PM, Pawel Jasinski wrote:
>>>>>> If you can find a nice Linux errno which maps from ERROR_SXS_CANT_GEN_ACTCTX
>>>>>> to something other than EACCES I'd be happy to change Cygwin.
>>> [...]
>>> Now I know that there are POSIX error codes and people tend to check
>>> error numbers and you want to put the closest match of a POSIX errno
>>> to the Windows error message. But why couldn't you just print to
>>> stderr the error message that Windows returned perhaps in addition
>>> to setting syserr to "Permission denied"?
>> Do you really want that?
>I think it would be helpful.
>> There's a difference between cmd and Cygwin.  Cmd is a shell, Cygwin is
>> just the underlying shared lib providing a generic API.
>OK so bash...

One of the points of Cygwin is that it allows you to build stuff like
bash with minimal or no modification.  That is the whole point of the
project.  If we have to modify Cygwin to provide a side-channel
communication mechanism for communicating Windows errors to a process
and then modify the process to use it that really flies in the face of
the project.

But you aren't really even suggesting that.  You are apparently just
suggesting that every windows error should be displayed by the Cygwin
DLL.  Wow.

>> If an error occurs, it's the shell's responsibility to print an error
>> message in the first place.  All messages printed by Cygwin are not
>> controllable by the calling application.  Therefore we usually only
>> print messages from the DLL if something very serious happens from the
>> DLLs perspective.  Some arbitrary Windows error code returned from
>> CreateProcess is usually not something actually serious.  There was
>> just "some" reason that an application couldn't be started.
>IMHO "some" reason that the user should be alerted about. How is it 
>helpful to the end user to suppress the error message?

There is no "suppression".  The error message that you see from cmd is
printed by cmd.  Cygwin isn't suppressing it.

>> Also, where do you draw the border?  Which windows error code is serious
>> enough to justify a (pretty intrusive!) error message from the underlying
>> library and which isn't?
>I would draw the border at "if there's an error message".

You clearly haven't thought this through.  Your screen would be filled with
errors.  What happens when something like bash looks for ~/.bashrc and it
doesn't exist?  You'd see a "File not found" on your screen.

Ok, so you don't want that one.  You just want the error messages that you
care about.  Which of the thousands of Windows errors would we decide to
display and which would we ignore?  It's not a solvable problem.

>> As cgf pointed out, Windows has zillions of error codes.  We wouldn't
>> want to generate the same number of POSIX-like error codes.  It wouldn't
>> make a lot of sense since POSIX applications only test for a limited,
>> expected number of error codes, and it might break things.
>
>I was talking error *messages* not error *codes*.

Cygwin and other windows programs do not see error *messages* when
something fails, they see error *codes*.  Getting the error *message*
involves translating the error *code* into a *message*.  We aren't actively
stopping you from seeing an error message.  We are translating a windows
error code into a POSIX errno.  Then bash reports the error using a table
of error strings that it gets from Cygwin.

The fact that this has to be explained to you pretty clearly illustrates that
you don't understand what is going on here.  You really should either go
thoroughly educate yourself about this or just stop talking.  Earnie's
points made here: http://cygwin.com/ml/cygwin/2012-08/msg00257.html are
quite pertinent.

cgf

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Andrew DeFaria]

On 08/13/2012 08:14 AM, Corinna Vinschen wrote:
>> There was an error message that cmd showed that bash did not. To me
>> that's suppression.
> This is nonsense and you know it.
It's not nonsense. It's 100% accurate. What in my statement above is not 
the truth?
>>> It's not that the OS generates an error message and cmd lets it slip
>>> through while Cygwin (or bash) "suppress" it.  It's the CreateProcess
>>> call which generates an error code ERROR_SXS_CANT_GEN_ACTCTX and cmd
>>> printing the connected error message, just like bash gets an error code
>>> EACCES and prints the connected error message "Permission denied".
>> Plumbing and mechanics aside, I'm just saying the user should be
>> told the underlying problem. If ERROR_SXS_CANT_GEN_ACTCTX is the
>> error code could ya at least print that as a string? It would give
>> the user a fighting change and finding a solution...
> Plumbing and mechanics is the way to solve the problem.
Right, that's makes it the implementors problem, plumbing and mechanics 
are implementation details - not the users...
> If you don't want to leave the user-only perspective, we will have a hard time to
> find a solution.
"We" are not finding a solution. "You" might be. "I'm" just reporting 
the problem. Now, does everybody have their roles straight? ;-)

Oh, and I fully expect it will not be solved. But I thought I'd report 
it nonetheless. My "side by side" problem has been fixed. I just think 
you should do a better job reporting more accurate error message so that 
users can more easily fix them. I didn't realize, but should have 
realized, that in this world it seems to be considered a crime...
-- 
Andrew DeFaria <http://defaria.com>
Do fish get cramps after eating?


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Andrew DeFaria]

On 08/13/2012 08:22 AM, Christopher Faylor wrote:
> But you aren't really even suggesting that.  You are apparently just
> suggesting that every windows error should be displayed by the Cygwin
> DLL.  Wow.
I never said "every" - you did.
>> I would draw the border at "if there's an error message".
> You clearly haven't thought this through.  Your screen would be filled with
> errors.  What happens when something like bash looks for ~/.bashrc and it
> doesn't exist?  You'd see a "File not found" on your screen.
That's not even a valid example. Even bash doesn't report that as an 
error as it isn't an error. You're being way to literal here. Let's 
change the statement "If there's an error, that you are gonna report, 
then you should report the best error that you can. One that leads to a 
solution the quickest". Call me silly but saying "Permission denied" in 
this circumstance is not "the best error you can report".

> Ok, so you don't want that one.  You just want the error messages that you
> care about.  Which of the thousands of Windows errors would we decide to
> display and which would we ignore?  It's not a solvable problem.
See above.
>
>>> As cgf pointed out, Windows has zillions of error codes.  We wouldn't
>>> want to generate the same number of POSIX-like error codes.  It wouldn't
>>> make a lot of sense since POSIX applications only test for a limited,
>>> expected number of error codes, and it might break things.
>> I was talking error *messages* not error *codes*.
> Cygwin and other windows programs do not see error *messages* when
> something fails, they see error *codes*.  Getting the error *message*
> involves translating the error *code* into a *message*.  We aren't actively
> stopping you from seeing an error message.  We are translating a windows
> error code into a POSIX errno.  Then bash reports the error using a table
> of error strings that it gets from Cygwin.
It's the "translation" that seems to be the problem. I understand, 
there's no error code that better translates. This, to me, just points 
out the flaw in the design of error handling and it results in bad error 
messages like the infamous "not a typewriter".
> The fact that this has to be explained to you pretty clearly illustrates that
> you don't understand what is going on here.
I do understand it. I gave a suggesting for how to better handle the 
error given the restrictions in the plumbing you describe.
> You really should either go thoroughly educate yourself about this or just stop talking.  Earnie's
> points made here: http://cygwin.com/ml/cygwin/2012-08/msg00257.html are
> quite pertinent.
Obviously you're not interested. So be it..
-- 
Andrew DeFaria <http://defaria.com>
Things are more like they are today than they ever were before.


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Corinna Vinschen]

On Aug 13 08:49, Andrew DeFaria wrote:
> On 08/13/2012 08:22 AM, Christopher Faylor wrote:
> >But you aren't really even suggesting that.  You are apparently just
> >suggesting that every windows error should be displayed by the Cygwin
> >DLL.  Wow.
> I never said "every" - you did.
> >>I would draw the border at "if there's an error message".
> >You clearly haven't thought this through.  Your screen would be filled with
> >errors.  What happens when something like bash looks for ~/.bashrc and it
> >doesn't exist?  You'd see a "File not found" on your screen.
> That's not even a valid example. Even bash doesn't report that as an
> error as it isn't an error. You're being way to literal here. Let's
> change the statement "If there's an error, that you are gonna
> report, then you should report the best error that you can. One that
> leads to a solution the quickest". Call me silly but saying
> "Permission denied" in this circumstance is not "the best error you
> can report".
> 
> >Ok, so you don't want that one.  You just want the error messages that you
> >care about.  Which of the thousands of Windows errors would we decide to
> >display and which would we ignore?  It's not a solvable problem.
> See above.
> >
> >>>As cgf pointed out, Windows has zillions of error codes.  We wouldn't
> >>>want to generate the same number of POSIX-like error codes.  It wouldn't
> >>>make a lot of sense since POSIX applications only test for a limited,
> >>>expected number of error codes, and it might break things.
> >>I was talking error *messages* not error *codes*.
> >Cygwin and other windows programs do not see error *messages* when
> >something fails, they see error *codes*.  Getting the error *message*
> >involves translating the error *code* into a *message*.  We aren't actively
> >stopping you from seeing an error message.  We are translating a windows
> >error code into a POSIX errno.  Then bash reports the error using a table
> >of error strings that it gets from Cygwin.
> It's the "translation" that seems to be the problem. I understand,
> there's no error code that better translates. This, to me, just
> points out the flaw in the design of error handling and it results
> in bad error messages like the infamous "not a typewriter".
> >The fact that this has to be explained to you pretty clearly illustrates that
> >you don't understand what is going on here.
> I do understand it. I gave a suggesting for how to better handle the
> error given the restrictions in the plumbing you describe.
> >You really should either go thoroughly educate yourself about this or just stop talking.  Earnie's
> >points made here: http://cygwin.com/ml/cygwin/2012-08/msg00257.html are
> >quite pertinent.
> Obviously you're not interested. So be it..

You can't be bothered to leave the high horse, right?  Did we really
sound like people *not* trying to solve this problem?  Unfortunately
not every solution is the best one, it always depends on the perspective
from where you're looking at it.  But you apparently don't want to see
that.  So, ok, probably you're right, we're obviously not interested.
So be it.


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader          cygwin AT cygwin DOT com
Red Hat

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Andrew DeFaria]

On 8/13/2012 9:01 AM, Corinna Vinschen wrote:
> You can't be bothered to leave the high horse, right? 
I like horses. Neigh, neigh as they say! :-)
> Did we really sound like people *not* trying to solve this problem? 
I didn't say that. All that remains is to see if you actually do solve 
the problem... I'm not holding my breath...
> Unfortunately not every solution is the best one, it always depends on 
> the perspective from where you're looking at it. But you apparently 
> don't want to see that. 
What so ever gave you that idea? I understand the perspective that you 
are seeing this from and I admit it makes it a tough nut to crack. Do 
you understand mine?
> So, ok, probably you're right, we're obviously not interested. So be 
> it. Corinna 
Cool, I'll drop out of this too.
-- 
Andrew DeFaria <http://defaria.com>
I used to have an open mind but my brains kept falling out.


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Andrey Repin]

Greetings, Christopher Faylor!

> But you aren't really even suggesting that.  You are apparently just
> suggesting that every windows error should be displayed by the Cygwin
> DLL.  Wow.

Um. On another hand, this made me think from my deep pit: as a regular user,
not a developer, in some cases, I would prefer DLL to track and report native
error codes/messages. Not as a general way of operation, obviously, more like
an intermediate debugging layer on demand.
So, there's a suggestion for someone interested in this feature and capable of
making this happen.


--
WBR,
Andrey Repin (anrdaemon@freemail.ru) 15.08.2012, <00:38>

Sorry for my terrible english...


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Christopher Faylor]

On Wed, Aug 15, 2012 at 12:42:09AM +0400, Andrey Repin wrote:
>Greetings, Christopher Faylor!

Sigh.

>>But you aren't really even suggesting that.  You are apparently just
>>suggesting that every windows error should be displayed by the Cygwin
>>DLL.  Wow.
>
>Um.  On another hand, this made me think from my deep pit: as a regular
>user, not a developer, in some cases, I would prefer DLL to track and
>report native error codes/messages.  Not as a general way of operation,
>obviously, more like an intermediate debugging layer on demand.  So,
>there's a suggestion for someone interested in this feature and capable
>of making this happen.

If you're "debugging" you can get the error by stepping into the DLL.

If you are actually capable of making it happen then prove it by
discussing implementation in cygwin-developers and eventually providing
a patch in cygwin-patches.  Otherwise, take Earnie's observation to
heart: http://cygwin.com/ml/cygwin/2012-08/msg00257.html

(I'm going to be quoting this a lot)

You aren't making any points by posting YA vague suggestion.

cgf

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Christopher Faylor]

On Tue, Aug 14, 2012 at 11:35:02PM +0200, Pawel Jasinski wrote:
>I am sorry I did not understand your comments.
>I am also not particularly familiar with your COP. I promise to adhere
>to it in the future.
>For the patches, I can only blame bad timing. I haven't done cvs up
>before posting.
>As you noticed for sure, I am also not so good with rhetoric.

Please honor my wishes.  There is quite obviously no reason for you to
send me private email about this.

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Pawel Jasinski]

Hi Christopher,

I am sorry I did not understand your comments.
I am also not particularly familiar with your COP. I promise to adhere
to it in the future.
For the patches, I can only blame bad timing. I haven't done cvs up
before posting.
As you noticed for sure, I am also not so good with rhetoric.

Cheers,
Pawel


On Tue, Aug 14, 2012 at 4:53 PM, Christopher Faylor
<cgf-use-the-mailinglist-pleasecygwin.com> wrote:
> On Tue, Aug 14, 2012 at 10:37:51AM +0200, Pawel Jasinski wrote:
>>Thanks guys!
>>That is a great news
>
> It isn't entirely clear that you understand that what just happened is
> what we always do when we find a Windows error which we haven't
> translated and for which there is better translation than EACCES.  I
> asked for suggestions, we got a good one, and I changed Cygwin.
>
> I wonder if much of this whole thread has just been a misconception
> about that SOP.
>
> cgf
>
> --
> Problem reports:       http://cygwin.com/problems.html
> FAQ:                   http://cygwin.com/faq/
> Documentation:         http://cygwin.com/docs.html
> Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
>

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Christopher Faylor]

On Tue, Aug 14, 2012 at 10:37:51AM +0200, Pawel Jasinski wrote:
>Thanks guys!
>That is a great news

It isn't entirely clear that you understand that what just happened is
what we always do when we find a Windows error which we haven't
translated and for which there is better translation than EACCES.  I
asked for suggestions, we got a good one, and I changed Cygwin.

I wonder if much of this whole thread has just been a misconception
about that SOP.

cgf

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Pawel Jasinski]

Thanks guys!
That is a great news

> cygwin1.dll returns ELIBADD natively (which, as
> of today's patches, it does)

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Eric Blake]

[-- Attachment #1: Type: text/plain, Size: 922 bytes --]

On 08/13/2012 06:07 PM, Pawel Jasinski wrote:
> hi,
> 
> How about this one. Bash has already some cygwin specific stuff.
> 
> rejap <at> win7dev /usr/src/bash-4.1.10-4/src/bash-4.1
> $ diff -uN execute_cmd.c.orig execute_cmd.c
> --- execute_cmd.c.orig  2012-08-14 00:36:57.092722900 +0200
> +++ execute_cmd.c       2012-08-14 01:50:03.248969800 +0200
> @@ -4838,6 +4838,10 @@
>             }
>  #endif
>           errno = i;
> +#if __CYGWIN__
> +      if (errno==EACCES && GetLastError() == ERROR_SXS_CANT_GEN_ACTCTX)
> +          errno = ELIBBAD;
> +#endif

Nope, won't do it.  If cygwin1.dll returns ELIBADD natively (which, as
of today's patches, it does), then bash will already do what you want
without me having to patch bash too.  Bash should not ever have to call
GetLastError().

-- 
Eric Blake   eblake@redhat.com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 620 bytes --]

Re: Side-by-side configuration is incorrect reported as permission denied

[Christopher Faylor]

On Tue, Aug 14, 2012 at 02:07:32AM +0200, Pawel Jasinski wrote:
>hi,
>
>How about this one. Bash has already some cygwin specific stuff.
>
>rejap <at> win7dev /usr/src/bash-4.1.10-4/src/bash-4.1
>$ diff -uN execute_cmd.c.orig execute_cmd.c
>--- execute_cmd.c.orig  2012-08-14 00:36:57.092722900 +0200
>+++ execute_cmd.c       2012-08-14 01:50:03.248969800 +0200
>@@ -4838,6 +4838,10 @@
>            }
> #endif
>          errno = i;
>+#if __CYGWIN__
>+      if (errno==EACCES && GetLastError() == ERROR_SXS_CANT_GEN_ACTCTX)
>+          errno = ELIBBAD;
>+#endif
>          file_error (command);
>        }
>       return ((i == ENOENT) ? EX_NOTFOUND : EX_NOEXEC);        /* XXX
>Posix.2 says that exit status is 126 */

Why would we do that instead of patching the DLL?  That doesn't make
any sense.

Are you somehow not getting the fact that the DLL was going to be
changed despite the fact that I said I was going to do it?  This has
been done.  You don't need to do anything here.

http://cygwin.com/ml/cygwin/2012-08/msg00253.html
http://cygwin.com/ml/cygwin/2012-08/msg00281.html
http://cygwin.com/ml/cygwin/2012-08/msg00302.html

Please drop this now.  It's clear that no one has any clever ideas about
how to make this work that doesn't involve patching every application
that uses errno.  So, this discussion should just die.

cgf

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Pawel Jasinski]

hi,

How about this one. Bash has already some cygwin specific stuff.

rejap <at> win7dev /usr/src/bash-4.1.10-4/src/bash-4.1
$ diff -uN execute_cmd.c.orig execute_cmd.c
--- execute_cmd.c.orig  2012-08-14 00:36:57.092722900 +0200
+++ execute_cmd.c       2012-08-14 01:50:03.248969800 +0200
@@ -4838,6 +4838,10 @@
            }
 #endif
          errno = i;
+#if __CYGWIN__
+      if (errno==EACCES && GetLastError() == ERROR_SXS_CANT_GEN_ACTCTX)
+          errno = ELIBBAD;
+#endif
          file_error (command);
        }
       return ((i == ENOENT) ? EX_NOTFOUND : EX_NOEXEC);        /* XXX
Posix.2 says that exit status is 126 */

Cheers,
Pawel

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Daniel Colascione]

[-- Attachment #1: Type: text/plain, Size: 1521 bytes --]

On 8/13/12 11:56 AM, Christopher Faylor wrote:
> On Mon, Aug 13, 2012 at 07:56:52PM +0200, Pawel Jasinski wrote:
>> hi guys,
>>
>> is that what we talking about (see patch below)?
> 
> Thanks for the patch but I wasn't looking for a simple patch to do this.
> I said I'd make the change if someone could suggest a better errno.
> 
> When I suggested a patch, it was to somehow change the way Cygwin
> reported errors so that Windows errors could be exposed to programs that
> aren't expecting them.  I was willing to look at an implementation if
> someone was going to take the time to add the non-trivial amount of code
> this would take.

I like the way the Windows CRT solves this problem. It provides a
traditional errno with the traditionally restricted set of error
codes, but also provides something called "_doserrno" that holds the
OS-level error code that generated the last errno. If Cygwin were to
provide a similar facility, Someone could patch high-visibility
programs like bash so that they'd display the Windows-level
information instead of the familiar strerror(errno) message. Granted,
not all errno-errors come from Win32 errors, so to avoid confusion,
you'd want to set _doserrno to zero every time you set errno, then (in
select places inside Cygwin) set _doserrno.

Looking at the bash code, I think the change there would involve
mechanically replacing all the uses of strerror (errno) with some
xstrerror (errno, _doserrno), and having xstrerror do the right thing.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 235 bytes --]

Re: Side-by-side configuration is incorrect reported as permission denied

[Christopher Faylor]

On Mon, Aug 13, 2012 at 07:56:52PM +0200, Pawel Jasinski wrote:
>hi guys,
>
>is that what we talking about (see patch below)?

Thanks for the patch but I wasn't looking for a simple patch to do this.
I said I'd make the change if someone could suggest a better errno.

When I suggested a patch, it was to somehow change the way Cygwin
reported errors so that Windows errors could be exposed to programs that
aren't expecting them.  I was willing to look at an implementation if
someone was going to take the time to add the non-trivial amount of code
this would take.

>If anybody is concerned about performance, I would be happy to rewrite
>lookup to use some sort of hashing

No thanks.

>I also tried to use koders.com to search for exec* and EACCES.
>It doesn't look like any code indexed by koders is fishing
>specifically for EACCES after call to exec* to do some magic recovery.

That really doesn't prove anything.

cgf

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Side-by-side configuration is incorrect reported as permission denied

[Pawel Jasinski]

hi guys,

is that what we talking about (see patch below)?
If anybody is concerned about performance, I would be happy to rewrite
lookup to use some sort of hashing

I also tried to use koders.com to search for exec* and EACCES.
It doesn't look like any code indexed by koders is fishing
specifically for EACCES after call to exec* to do some magic recovery.


--pawel

Index: newlib/libc/include/sys/errno.h
===================================================================
RCS file: /cvs/src/src/newlib/libc/include/sys/errno.h,v
retrieving revision 1.18
diff -u -r1.18 errno.h
--- newlib/libc/include/sys/errno.h     26 Dec 2010 21:11:38 -0000      1.18
+++ newlib/libc/include/sys/errno.h     13 Aug 2012 17:48:33 -0000
@@ -116,7 +116,9 @@
 #define EBADFD 81      /* f.d. invalid for this operation */
 #define EREMCHG 82     /* Remote address changed */
 #define ELIBACC 83     /* Can't access a needed shared lib */
+#endif
 #define ELIBBAD 84     /* Accessing a corrupted shared lib */
+#ifdef __LINUX_ERRNO_EXTENSIONS__
 #define ELIBSCN 85     /* .lib section in a.out corrupted */
 #define ELIBMAX 86     /* Attempting to link in too many libs */
 #define ELIBEXEC 87    /* Attempting to exec a shared library */

Index: winsup/cygwin/errno.cc
===================================================================
RCS file: /cvs/src/src/winsup/cygwin/errno.cc,v
retrieving revision 1.87
diff -u -r1.87 errno.cc
--- winsup/cygwin/errno.cc      3 Dec 2011 21:43:25 -0000       1.87
+++ winsup/cygwin/errno.cc      13 Aug 2012 17:48:44 -0000
@@ -157,6 +157,7 @@
   X (WORKING_SET_QUOTA,                EAGAIN),
   X (WRITE_PROTECT,            EROFS),
   X (PRIVILEGE_NOT_HELD,       EPERM),
+  X (SXS_CANT_GEN_ACTCTX, ELIBBAD),
   { 0, NULL, 0}
 };

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple