public inbox for cygwin@cygwin.com
 help / color / mirror / Atom feed
* Re: cygwin bughunt (FAQ alert?)
@ 2005-01-26 15:19 David Dindorp
  2005-01-27  1:56 ` Joshua Daniel Franklin
  0 siblings, 1 reply; 39+ messages in thread
From: David Dindorp @ 2005-01-26 15:19 UTC (permalink / raw)
  To: cygwin


Igor Pechtchanski wrote:
>> Umm, that was my bad.  The thing is, "--enable-debugging" really
produces
>> a developer debug version, with extra tracing, etc.  If all you want
is a
>> version of DLL with all the symbols (i.e., unstripped), the regular
build
>> produces that as well.

Cristopher Faylor wrote:
> ...and now you get to repeat these facts endlessly as people find your
> words in the archives and assume that they need use this option
regardless
> of follow-on discussion or the FAQ.

How about adding a line in the FAQ to the "how to build cygwin" (104)
entry
stating that the "configure ; make" mentioned does produce a Cygwin with
all
debugging symbols?

And the link in the FAQ is wrong:

"How can I debug cygwin" (entry 105) says:

"To build a debugging version of the Cygwin DLL,
 you will need to follow the instructions at
 http://cygwin.com/faq/faq_3.html#SEC102."

The above should point to entry 104, not 102.



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

^ permalink raw reply	[flat|nested] 39+ messages in thread
* Re: cygwin bughunt (FAQ alert?)
@ 2005-01-26 16:35 David Dindorp
  0 siblings, 0 replies; 39+ messages in thread
From: David Dindorp @ 2005-01-26 16:35 UTC (permalink / raw)
  To: cygwin

Ack!
Apologies for the formatting.
The company I'm employed at uses Outlook (thereby MS-WORD) for e-mail.

Here's what I wanted to say:

The FAQ entry 105 links to entry 102 under "how to compile".
Shouldn't this point to 104 instead?



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

^ permalink raw reply	[flat|nested] 39+ messages in thread
* Re: cygwin bughunt (FAQ alert?)
@ 2005-01-25 20:10 David Dindorp
  2005-01-25 21:59 ` Igor Pechtchanski
  0 siblings, 1 reply; 39+ messages in thread
From: David Dindorp @ 2005-01-25 20:10 UTC (permalink / raw)
  To: cygwin

Cristopher Faylor wrote:
> Actually, we do.  We provide the source code.  It's easy to build.

On your particular system which is tuned to do precisely this, maybe.

If it's as easy as you say, I'll spend some more time on it.


> Have you even tried it?

No.  For a couple of reasons.

1. Prior experience with compiling large open source projects have
shown me that it usually takes intimate knowledge of the source code
and what tools and operating system should be used in order to make
a good compile.

2. Other peoples posting suggested compiling Cygwin is not a walk in
the park.

3. I had no idea of the "--enable-debugging" option that creates a
debug version, or any of the other requirements for making the source
compile (I'm sure there exists some).

4. I hate to bug the list with stupid questions on how to compile
Cygwin, when all I really need is to retrieve more debug information
from a running system, not compile a new one.

5.  Probably more reasons.  Nobody cares, so I'm going to stop the
listing here :-).


I just tried a regular (non-debug) compile, compiling the freshest
source that comes with the stock 1.5.10, using GCC etc. from 1.5.10.
It stopped compiling with this error message:

/winsup/cygwin/errno.cc:154: error:
 external linkage required for symbol 'const char* const _sys_errlist[]'
 because of 'dllexport' attribute.

The cause for this particular compile error is probably some minor
technicality, but add up a dozen of these, and I will soon have spent a
month just trying to make myself a debug DLL to match my 1.5.10 :-(.

I've seen advice elsewhere to simply migrate through Cygwin versions
until I happen to bump into one that works with the application in
question.  With limited time on my hands (I can't devote a month to
finding out how to compile Cygwin proper), I think that that's maybe
the most viable solution so far?..

Otherwise I may put some more effort into the whole compilation thing.
There's another version of the source that comes with stock 1.5.10
which so far only complains about missing 'w32api'-something, so maybe
I'll have more luck with that.



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

^ permalink raw reply	[flat|nested] 39+ messages in thread
* Re: cygwin bughunt (FAQ alert?)
@ 2005-01-21 18:49 David Dindorp
  2005-01-22 12:18 ` Christopher Faylor
  0 siblings, 1 reply; 39+ messages in thread
From: David Dindorp @ 2005-01-21 18:49 UTC (permalink / raw)
  To: cygwin

Corinna Vinschen wrote:
> IMHO you're looking from the wrong direction.  People capable of
> debugging the Cygwin DLL are usually also capable of building it.

The only reason that the above is true is because you do not provide
the means for people to debug the Cygwin DLL properly.

> I'm wondering how somebody should be able to debug an application
> at all, if this person stumbles over using the compiler tools.

In the real world there is no strong binding between being able to
compile a properly functioning Cygwin DLL, and being able to look
through the source code, follow the developer's chain of thought and
figuring out why things do not work given the appropriate debug
information.  You imply that in order to compile a working Cygwin,
an intelligence quotient of X is required, while in order to debug it,
a higher intelligence quotient X + Y is required.
That's just not true.

Entirely different sets of skills are involved.

I'll admit though that being able to compile a functioning Cygwin
makes debugging easier by removing a lot of the brain work required,
and replacing it with simple trial-and-error.

That approach is unfortunately just plain impossible
when it comes to race conditions (eg.) or production systems.



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

^ permalink raw reply	[flat|nested] 39+ messages in thread
* RE: cygwin bughunt (FAQ alert?)
@ 2005-01-21 14:59 Hughes, Bill
  0 siblings, 0 replies; 39+ messages in thread
From: Hughes, Bill @ 2005-01-21 14:59 UTC (permalink / raw)
  To: 'cygwin@cygwin.com'

Corinna Vinschen wrote:
> On Jan 21 11:18, Hughes, Bill wrote:
>> I don't think I'm putting this very well, but it may make the FAQ
>> easier if the standard advice is to load the snaphot and use that
>> for debugging, it removes a separate layer of potential problems in
>> building the dll. I suspect the people who would want a stripped
>> snapshot to be more capable of producing it than those would may
>> need to build one with debug info. 
> 
> IMHO you're looking from the wrong direction.  People capable
> of debugging
> the Cygwin DLL are usually also capable of building it.  I'm wondering
> how somebody should be able to debug an application at all,
> if this person
> stumbles over using the compiler tools.
Which is why I asked, I suspect I was hoping there was a way to help newbies
(like me in this respect) to generate useful reports in cases of suspected
bugs for the more knowledgeable to read.
Of course if there were such a way I would expect someone else to have
thought of it, and so it's probably impracticable.

Thanks for the reply,
Bill
-- 
       ___
  oo  // \\      "De Chelonian Mobile"
 (_,\/ \_/ \     TortoiseSVN
   \ \_/_\_/>    The coolest Interface to (Sub)Version Control
   /_/   \_\     http://tortoisesvn.tigris.org

This e-mail transmission is strictly confidential and intended solely
for the person or organisation to whom it is addressed. It may contain
privileged and confidential information and if you are not the intended
recipient, you must not copy, distribute or take any action in reliance
on it. If you have received this email in error, please reply to the
sender as soon as possible and delete the message. Please note that we
are able to, and reserve the right to, monitor e-mail communications
passing through our network.

The views expressed in this email are not that of the company unless
specified within the message.

The inclusion of this footnote indicates that the mail message and any
attachments have been checked for the presence of known viruses.

If you have any comments regarding our policy please direct them to
postmaster@cox.co.uk
________________________________________________________________________
This email has been scanned for all viruses by the MessageLabs Email
Security System. For more information on a proactive email security
service working around the clock, around the globe, visit
http://www.messagelabs.com
________________________________________________________________________

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

^ permalink raw reply	[flat|nested] 39+ messages in thread
* RE: cygwin bughunt (FAQ alert?)
@ 2005-01-21 12:04 Hughes, Bill
  2005-01-21 12:56 ` Corinna Vinschen
  0 siblings, 1 reply; 39+ messages in thread
From: Hughes, Bill @ 2005-01-21 12:04 UTC (permalink / raw)
  To: 'cygwin@cygwin.com'

Christopher Faylor wrote:
..snip..
>> The snapshots page says that it's a stripped version.
>> Who should I trust, the snapshot page or the FAQ?
> 
> You should trust me when I tell you that the snapshots haven't been
> stripped recently. 
> 
> However, oops, this means that the advice of using a snapshot
> shouldn't go into the FAQ since this isn't a permanent arrangement.

Out of curiosity, would it make sense to always build the snapshot with the
debug info?
Thinking about the 'hierarchy of ignorance' for want of a better term, does
it require more knowledge to run gdb and give a sensible report or to build
the dll and then do the same?
I don't think I'm putting this very well, but it may make the FAQ easier if
the standard advice is to load the snaphot and use that for debugging, it
removes a separate layer of potential problems in building the dll. I
suspect the people who would want a stripped snapshot to be more capable of
producing it than those would may need to build one with debug info.

Bill
-- 
       ___
  oo  // \\      "De Chelonian Mobile"
 (_,\/ \_/ \     TortoiseSVN
   \ \_/_\_/>    The coolest Interface to (Sub)Version Control
   /_/   \_\     http://tortoisesvn.tigris.org

This e-mail transmission is strictly confidential and intended solely
for the person or organisation to whom it is addressed. It may contain
privileged and confidential information and if you are not the intended
recipient, you must not copy, distribute or take any action in reliance
on it. If you have received this email in error, please reply to the
sender as soon as possible and delete the message. Please note that we
are able to, and reserve the right to, monitor e-mail communications
passing through our network.

The views expressed in this email are not that of the company unless
specified within the message.

The inclusion of this footnote indicates that the mail message and any
attachments have been checked for the presence of known viruses.

If you have any comments regarding our policy please direct them to
postmaster@cox.co.uk
________________________________________________________________________
This email has been scanned for all viruses by the MessageLabs Email
Security System. For more information on a proactive email security
service working around the clock, around the globe, visit
http://www.messagelabs.com
________________________________________________________________________

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

^ permalink raw reply	[flat|nested] 39+ messages in thread
* Re: cygwin bughunt (FAQ alert?)
@ 2005-01-21  0:39 David Dindorp
  2005-01-21  1:01 ` Christopher Faylor
  0 siblings, 1 reply; 39+ messages in thread
From: David Dindorp @ 2005-01-21  0:39 UTC (permalink / raw)
  To: Cygwin List

David Dindorp wrote:
> Tracking it down with GDB to cygwin_split_path() : 0x61073e06 was
easy.

Christopher Faylor wrote:
> Since cygwin isn't built with debugging symbols, the symbols that you
do
> see in gdb are basically meaningless.

Isn't there any way to compile the debugging symbols into a separate
file
that GDB could then play with if it wanted to?


Joshua Daniel Franklin wrote:
>> I think this has come up often enough to be a FAQ.  Joshua?
> Sure, how about this:

> either build your own debugging version by following the instructions
> at http://cygwin.com/faq/faq_3.html#SEC102 or use a current snapshot
> from http://cygwin.com/snapshots/

Eep.  Totally overlooked the entire FAQ.  I apologize!!!

The snapshots page says that it's a stripped version.
Who should I trust, the snapshot page or the FAQ?

Is it considered atrocious to just replace the DLL with a snapshot one
and keep the EXE's from stock?


Pierre A. Humblet wrote:
> Should we also provide an optional cygwin_debug package, with only
> an unstripped cygwin1.dll.debug ?

I for one would be eternally grateful :-).
FWIW, MySQL AB does the same with MyODBC (in the same 'package' though).
I've found it useful a number of times.


'Arigato gozaimasu' *bows*.
--david



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

^ permalink raw reply	[flat|nested] 39+ messages in thread
* Re: cygwin bughunt
@ 2005-01-20 20:05 David Dindorp
  2005-01-20 20:47 ` cygwin bughunt (FAQ alert?) Christopher Faylor
  0 siblings, 1 reply; 39+ messages in thread
From: David Dindorp @ 2005-01-20 20:05 UTC (permalink / raw)
  To: Cygwin List

Larry Hall wrote:

> I have the following suggestions/questions:

>  1. Did you try a Cygwin 1.5.12 or even a snapshot?

No.  I'm using 1.5.10, and it still smells *real* fresh, I think ;-).

Also, the problem only occurs on a customer system which unfortunately
I can't go around and upgrade all the time "just to see".
I don't have regular access to it either.


>  2. Is this a local debug build of Cygwin or stock 1.5.10.  If the 
>     latter, you might find building a debug version is more help.

Stock 1.5.10.
There's a debug version?  These are the things I need to know!
Found the "how-to-debug-cygwin" document buried deep in some bz2 file,
I'll go read that now.

In the above statement, you sort-of imply that building a debug version
is wrong if using a stock version..  I should do what then?


> If there is a race issue here, you're going to need to work with the 
> code to find it.

That's what I'm basically trying to do.
Tracking it down with GDB to cygwin_split_path() : 0x61073e06 was easy.
But then I'm kinda stuck.

Stuff that would be REALLY nice to know would for example be:
 - How do I find the call parameters to this function when it died?
   I poked around all the "info locals/variables" etc. commands in GDB
   and they basically just tell me that there's none to be found.
 - How do I translate "cygwin_split_path() : 0x61073e06" to a source
   file name and a line number?


Thank you for the suggestions (please bring more ;-))!
/david



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

^ permalink raw reply	[flat|nested] 39+ messages in thread
end of thread, other threads:[~2005-01-27  4:16 UTC | newest]

Thread overview: 39+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2005-01-26 15:19 cygwin bughunt (FAQ alert?) David Dindorp
2005-01-27  1:56 ` Joshua Daniel Franklin
2005-01-27  4:51   ` Joshua Daniel Franklin
2005-01-27  6:38     ` Brian Dessent
  -- strict thread matches above, loose matches on Subject: below --
2005-01-26 16:35 David Dindorp
2005-01-25 20:10 David Dindorp
2005-01-25 21:59 ` Igor Pechtchanski
2005-01-25 22:54   ` Christopher Faylor
2005-01-21 18:49 David Dindorp
2005-01-22 12:18 ` Christopher Faylor
2005-01-21 14:59 Hughes, Bill
2005-01-21 12:04 Hughes, Bill
2005-01-21 12:56 ` Corinna Vinschen
2005-01-21 16:57   ` Christopher Faylor
2005-01-21 17:00     ` Igor Pechtchanski
2005-01-21 18:21       ` Corinna Vinschen
2005-01-21 19:26       ` Christopher Faylor
2005-01-21 19:42         ` Igor Pechtchanski
2005-01-21 21:54           ` Christopher Faylor
2005-01-21 19:47         ` Pierre A. Humblet
2005-01-21 22:00           ` Christopher Faylor
2005-01-21 23:11             ` Pierre A. Humblet
2005-01-22  4:26               ` Christopher Faylor
2005-01-22 21:00                 ` Joshua Daniel Franklin
2005-01-22 22:23                   ` Christopher Faylor
2005-01-22 23:44                     ` Joshua Daniel Franklin
2005-01-23  1:00                       ` Christopher Faylor
2005-01-24  6:52                         ` Joshua Daniel Franklin
2005-01-21  0:39 David Dindorp
2005-01-21  1:01 ` Christopher Faylor
2005-01-21  8:30   ` Joshua Daniel Franklin
2005-01-20 20:05 cygwin bughunt David Dindorp
2005-01-20 20:47 ` cygwin bughunt (FAQ alert?) Christopher Faylor
2005-01-20 21:21   ` Joshua Daniel Franklin
2005-01-20 21:35     ` Christopher Faylor
2005-01-20 21:38     ` Igor Pechtchanski
2005-01-20 21:47       ` Christopher Faylor
2005-01-20 22:35     ` Pierre A. Humblet
2005-01-21 11:57       ` Corinna Vinschen
2005-01-21 15:59         ` Pierre A. Humblet

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).