[ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Eric Blake (cygwin)]

The releases of readline 7.0.1-1 and bash 4.4.5-1, which have been
experimental for a few weeks, have now been promoted to current.  Please
refer to the earlier announcements for more information about these
releases:

https://cygwin.com/ml/cygwin-announce/2016-12/msg00009.html
https://cygwin.com/ml/cygwin-announce/2016-12/msg00028.html

QUESTIONS:
==========
If you want to make a point or ask a question the Cygwin mailing list is
the appropriate place.

-- 
Eric Blake
volunteer cygwin readline/bash package maintainer

For more details on this list (including unsubscription), see:
http://sourceware.org/lists.html

--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Tue, 3 Jan 2017 10:34:01, "Eric Blake (cygwin)" wrote:
> The releases of readline 7.0.1-1 and bash 4.4.5-1, which have been
> experimental for a few weeks, have now been promoted to current.

Note that new version of libreadline7 breaks interactive non ASCII input. Please
see:

http://cygwin.com/ml/cygwin/2017-01/msg00066.html

downgrading to the following versions fixes the problem:

- libreadline7-6.3.8
- bash-4.3.48


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Eric Blake]

[-- Attachment #1.1: Type: text/plain, Size: 1006 bytes --]

On 01/12/2017 11:05 PM, Steven Penny wrote:
> On Tue, 3 Jan 2017 10:34:01, "Eric Blake (cygwin)" wrote:
>> The releases of readline 7.0.1-1 and bash 4.4.5-1, which have been
>> experimental for a few weeks, have now been promoted to current.
> 
> Note that new version of libreadline7 breaks interactive non ASCII input. Please
> see:
> 
> http://cygwin.com/ml/cygwin/2017-01/msg00066.html
> 
> downgrading to the following versions fixes the problem:
> 
> - libreadline7-6.3.8
> - bash-4.3.48

Is the problem in libreadline7 or in bash? The newer bash requires the
newer readline, but if the problem is in readline proper, then the newer
readline plus the older bash would reproduce it.

I don't think I introduced any accidental downstream regressions, so
this may be a true upstream regression in readline.  But I have not yet
had time to try and reproduce the problem.

-- 
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: 604 bytes --]

Re: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Fri, 13 Jan 2017 08:46:43, Eric Blake wrote:
> Is the problem in libreadline7 or in bash? The newer bash requires the
> newer readline, but if the problem is in readline proper, then the newer
> readline plus the older bash would reproduce it.

1. new readline, new bash = bad

2. new readline, old bash = bad

3. old readline, new bash:

> Entry Point Not Found
> The procedure entry point history_lines_read_from_file could not be located in
> the dynamic link library cygreadline7.dll.

4. old readline, old bash = good


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Eric Blake]

[-- Attachment #1.1: Type: text/plain, Size: 812 bytes --]

On 01/13/2017 06:23 PM, Steven Penny wrote:
> On Fri, 13 Jan 2017 08:46:43, Eric Blake wrote:
>> Is the problem in libreadline7 or in bash? The newer bash requires the
>> newer readline, but if the problem is in readline proper, then the newer
>> readline plus the older bash would reproduce it.
> 
> 1. new readline, new bash = bad
> 
> 2. new readline, old bash = bad

Okay, so you've isolated the problem to readline.  In all likelihood, it
is an unintentional upstream regression; now we need to figure out what
is causing the change in behavior.

> 
> 3. old readline, new bash:
> 
>> Entry Point Not Found

Expected. New bash requires the new readline (as I already stated).

-- 
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: 604 bytes --]

Re: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Sat, 14 Jan 2017 15:47:57, Eric Blake wrote:
> Okay, so you've isolated the problem to readline.  In all likelihood, it
> is an unintentional upstream regression; now we need to figure out what
> is causing the change in behavior.

I would like to make it very clear that I will not be digging around in the
readline source code to fix this. I think I have gone above and beyond in
diagnosing this for you, so if you can take over it would be much appreciated.


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Vince Rice]

> On Jan 14, 2017, at 6:27 PM, Steven Penny <svnpenn@gmail.com> wrote:
> 
> On Sat, 14 Jan 2017 15:47:57, Eric Blake wrote:
>> Okay, so you've isolated the problem to readline.  In all likelihood, it
>> is an unintentional upstream regression; now we need to figure out what
>> is causing the change in behavior.
> 
> I would like to make it very clear that I will not be digging around in the
> readline source code to fix this. I think I have gone above and beyond in
> diagnosing this for you, so if you can take over it would be much appreciated.

It's times like this I wish CGF were still here…
--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[cyg Simple]

On 1/15/2017 8:23 AM, Vince Rice wrote:
>> On Jan 14, 2017, at 6:27 PM, Steven Penny <svnpenn@gmail.com> wrote:
>>
>> On Sat, 14 Jan 2017 15:47:57, Eric Blake wrote:
>>> Okay, so you've isolated the problem to readline.  In all likelihood, it
>>> is an unintentional upstream regression; now we need to figure out what
>>> is causing the change in behavior.
>>
>> I would like to make it very clear that I will not be digging around in the
>> readline source code to fix this. I think I have gone above and beyond in
>> diagnosing this for you, so if you can take over it would be much appreciated.
> 
> It's times like this I wish CGF were still here…

Yes, being blunt about open source is everyone's responsibility,
including Steven Penny's, was a knack he had down pat.  Steven is just
plain rude with his expectations of others free time and his
paid-for-support model of expectations.  Steven, bullying your way
around this list is likely to get you band from it and it certainly
isn't appreciated that you *expect* others to do work you're not willing
to do.  It is this expectation that others do the work that you freely
use that is the problem.  So let's make it very clear we will not bend
over backward to resolve issues you *demand* to be fixed without
acknowledgement of others free time to do so.  Willingness to help is
the best thing you can offer even if you must decline to do so.

-- 
cyg 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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Sun, 15 Jan 2017 09:11:23, cyg Simple wrote:
> It is this expectation that others do the work that you freely use that is the
> problem.

I dont think you realize, I am the maintainer of the Cygwin package manager:

http://github.com/svnpenn/sage

to which I have put hundreds of hours of work into, free of charge. In addition
I donate over a 1,000 commits a year to open source:

http://github.com/svnpenn

And have helped millions of people on Stack Overflow:

http://stackoverflow.com/users/1002260/steven-penny

So please, dont tell me that I am not willing to do open source work. Yes, I do
expect Eric to fix this. readline is a critial Cygwin package and he let a
pretty glaring error get though. It is scary to think what testing he even did
before releasing it. The least he could do is roll it back until the problem has
been fixed by himself or upstream.


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Thomas Wolff]

Am 14.01.2017 um 22:47 schrieb Eric Blake:
> On 01/13/2017 06:23 PM, Steven Penny wrote:
>> On Fri, 13 Jan 2017 08:46:43, Eric Blake wrote:
>>> Is the problem in libreadline7 or in bash? ...
> Okay, so you've isolated the problem to readline.  In all likelihood, it
> is an unintentional upstream regression; now we need to figure out what
> is causing the change in behavior.
Or actually, whether the current (now broken) functionality is useful at 
all, as it does not insert the character with the code being entered...
In a UTF-8 locale, e.g., Alt+2-1-6 should result in the Unicode 
character with decimal code 216, which is 'Ø', and so does mintty (with 
or without NumLock, by the way) – but it inserts 'Ï' which has code 216 
in the ancient PC/VGA character set (CP437). I think this is pretty useless.
I also think it would be better to place this functionality into the 
console handler in the cygwin dll on that occasion, and clear it from 
readline. If there is agreement on this, I might eventually look into 
the implementation...
Thomas

--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Sun, 15 Jan 2017 23:33:48, Thomas Wolff wrote:
> In a UTF-8 locale, e.g., Alt+2-1-6 should result in the Unicode 
> character with decimal code 216

No, it shouldnt. It should result in U+256A.

http://wikipedia.org/wiki/Code_page_437

Please advise where you are getting your information? Also even if the wrong
code page is being used, new version of readline still broke pasting, which will
still need to be fixed regardless of any Alt code situation.


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Sun, 15 Jan 2017 16:44:57, Steven Penny wrote:
> On Sun, 15 Jan 2017 23:33:48, Thomas Wolff wrote:
> > In a UTF-8 locale, e.g., Alt+2-1-6 should result in the Unicode 
> > character with decimal code 216

Actually, this does work:

- Alt 216 gives you U+256A via code page 437

- Alt 0216 gives you U+00D8 via UTF-8

http://fileformat.info/tip/microsoft/enter_unicode.htm


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Yaakov Selkowitz]

On 2017-01-15 08:30, Steven Penny wrote:
> On Sun, 15 Jan 2017 09:11:23, cyg Simple wrote:
>> It is this expectation that others do the work that you freely use that is the
>> problem.
>
> I dont think you realize, I am the maintainer of the Cygwin package manager:
>
> http://github.com/svnpenn/sage

Whatever that is, it is NOT *the* Cygwin package manager.

> to which I have put hundreds of hours of work into, free of charge. In addition
> I donate over a 1,000 commits a year to open source:
>
> And have helped millions of people on Stack Overflow:

None of which is relevant to the point that...

> So please, dont tell me that I am not willing to do open source work. Yes, I do
> expect Eric to fix this.

This still does not give you the right to make demands.  Remember that.

> readline is a critial Cygwin package and he let a
> pretty glaring error get though. It is scary to think what testing he even did
> before releasing it. The least he could do is roll it back until the problem has
> been fixed by himself or upstream.

Eric has been around for a long time and has earned the respect of this 
community.  He knows what his responsibilities are and will surely 
address them as always.

-- 
Yaakov

--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Mon, 16 Jan 2017 04:32:21, Yaakov Selkowitz wrote:
> None of which is relevant to the point

It certainly is. cyg Simple said:

> you *expect* others to do work you're not willing to do

I demonstrated that I am willing to do said work.

> This still does not give you the right to make demands.  Remember that.

Thank you for clarifiying, but what demand did I make? Certainly I do expect
Eric to fix this, but I am not demanding it of him? I he doesnt fix it, I will
just continue to use the old version as I am doing now.


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Sat, 14 Jan 2017 15:47:57, Eric Blake wrote:
> Okay, so you've isolated the problem to readline.  In all likelihood, it
> is an unintentional upstream regression; now we need to figure out what
> is causing the change in behavior.

Git Bisect reveals this:

    d49a9082c0e15bba8cd3d8cc0a994409cf823cac is the first bad commit
    commit d49a9082c0e15bba8cd3d8cc0a994409cf823cac
    Author: Chet Ramey <chet.ramey@case.edu>
    Date:   Thu Sep 15 17:02:15 2016 -0400

        Readline-7.0 distribution sources and documentation

Although that is not very helpful because d49a908 is -16034/+20692:

http://git.savannah.gnu.org/cgit/readline.git/commit?id=d49a908

Basically they dumped the whole thing into 1 commit. Kind of pointless to even
use Git when you are doing to do that. Thanks a lot Chet :|


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Mon, 16 Jan 2017 18:47:14, Steven Penny wrote:
> Although that is not very helpful because d49a908 is -16034/+20692:

I did another bisect on the readline-7.0-testing branch and came up with this:

    ef27d114778ffef483ed2746603f9487f880edbd is the first bad commit
    commit ef27d114778ffef483ed2746603f9487f880edbd
    Author: Chet Ramey <chet.ramey@case.edu>
    Date:   Thu Jul 9 16:28:21 2015 -0400

        readline-7.0-alpha distribution

at -9728/+13220, it is still not super helpful, but better than the previous
bisect.


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Mon, 16 Jan 2017 19:32:19, Steven Penny wrote:
> I did another bisect on the readline-7.0-testing branch and came up with this:
>     ef27d114778ffef483ed2746603f9487f880edbd is the first bad commit
>     commit ef27d114778ffef483ed2746603f9487f880edbd
>     Author: Chet Ramey <chet.ramey@case.edu>
>     Date:   Thu Jul 9 16:28:21 2015 -0400
>         readline-7.0-alpha distribution

On Linux readline is 2 packages:

libreadline7 = 7.0
libreadline6 = 6.3

Using:

- Debian 8.7.1
- http://packages.debian.org/stretch/libreadline7

For the test in question, GREEK CAPITAL LETTER OMEGA U+03A9, you do:

Ctrl + Shift + U, release, 3a9, Enter

This still works on Linux, even with libreadline7. Chet is refusing to address
the issue:

http://lists.gnu.org/archive/html/bug-readline/2017-01/msg00001.html

So it looks like we will need a Cygwin specific patch?


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Eric Blake]

[-- Attachment #1.1: Type: text/plain, Size: 1729 bytes --]

[your email client tends to break threading, which is making it VERY
hard to follow your multiple replies]

On 01/13/2017 08:46 AM, Eric Blake wrote:
> On 01/12/2017 11:05 PM, Steven Penny wrote:
>> On Tue, 3 Jan 2017 10:34:01, "Eric Blake (cygwin)" wrote:
>>> The releases of readline 7.0.1-1 and bash 4.4.5-1, which have been
>>> experimental for a few weeks, have now been promoted to current.
>>
>> Note that new version of libreadline7 breaks interactive non ASCII input. Please
>> see:
>>
>> http://cygwin.com/ml/cygwin/2017-01/msg00066.html
>>
>> downgrading to the following versions fixes the problem:
>>
>> - libreadline7-6.3.8
>> - bash-4.3.48

So far, I'm unable to reproduce your issue.  Here's what I've tried:

In a clean windows VM, I've opened up cmd.exe, and verified that
ALT-2-3-4 generates 'Ω' (so my default locale encodings and font in that
screen do what you describe).  Likewise, ALT-6-6 generates 'B'.  From
there, I created a test directory with JUST the following files:

bash.exe from bash-4.3.48-8
cyg{history,readline7}.dll from libreadline-6.3.8-1
cyg{gcc_s-1,iconv-2,intl-8,ncursesw-10,win1}.dll from current setup.exe

and verified that it was sufficient to start up the older bash in the
same cmd.exe window (invoking .\bash).  Once that version of bash is up
and running, ALT-6-6 types 'B', but ALT-2-3-4 types nothing.

Based on your description, you said the older combination lets things
through.  Are you sure you didn't also downgrade cygwin1.dll?  At this
point, I'm inclined to think the problem is in cygwin1.dll and not
libreadline7.

-- 
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: 604 bytes --]

Re: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Eric Blake]

[-- Attachment #1.1: Type: text/plain, Size: 3057 bytes --]

[again, your email client likes to break threading, so I nearly missed
this one]

On 01/17/2017 06:36 PM, Steven Penny wrote:
> On Mon, 16 Jan 2017 19:32:19, Steven Penny wrote:
>> I did another bisect on the readline-7.0-testing branch and came up with this:
>>     ef27d114778ffef483ed2746603f9487f880edbd is the first bad commit
>>     commit ef27d114778ffef483ed2746603f9487f880edbd
>>     Author: Chet Ramey <chet.ramey@case.edu>
>>     Date:   Thu Jul 9 16:28:21 2015 -0400
>>         readline-7.0-alpha distribution
> 

Are you building readline from source each step of the bisection? And
are you applying all the downstream cygwin patches each build, or just
using the stock upstream sources? (The cygwin patches can be found in
/usr/src if you use setup.exe to download the sources)

> This still works on Linux, even with libreadline7.

Linux doesn't have a cmd.exe.  And in mintty, I'm having no problems
entering extended characters (well, there, ALT-6-6 gives 'B' as in
ASCII, but ALT-2-3-4 gives 'ê' instead of 'Ω', while ALT-+-2-3-4 gives
'Ω', but at least input isn't silently dropped).  So I really think this
is an issue with interacting with the native windows console (which is
cygwin's domain), and not with the input being fed to readline (readline
is reading the terminal presented by cygwin, but it is cygwin that is
interpreting the console's reaction to keypresses in a way that is
converted to what cygwin apps think is a normal terminal).

Another thing I tried: running bash in mintty, this command:
 printf $'\u66\u3a9'
outputs 'fΩ' (correct, since hex 66 is 'f', while decimal 66 is 'B').
But while running bash directly from cmd.exe, the same command results
in 'f\u03A9'.  It's as if bash has determined (from locale?) that the
current character set is not Unicode, and therefore outputs an escape
sequence instead of the UTF-8 sequence.  So next I tried
LANG=en_US.UTF-8. After that, the printf produced the results I got in
mintty, but ALT-2-3-4 caused \316 to be output.

Normally, I _don't_ run anything under cmd.exe (all my use of cygwin is
under mintty), so I need a lot of hand-holding to reproduce anything in
your setup, in particular how you are even getting extended characters
to print (what locales and/or window settings you have to set).

> Chet is refusing to address
> the issue:
> 
> http://lists.gnu.org/archive/html/bug-readline/2017-01/msg00001.html

I didn't even know bug-readline existed (all the readline bug reports
I've seen over the years have been addressed to bug-bash); I'm
subscribed now.  But Chet is right that it is probably a Cygwin problem,
and not a readline problem.

> 
> So it looks like we will need a Cygwin specific patch?

Well, first someone has to get to the real root cause. Since I have been
unable to reproduce it, and I don't have much free time to devote to it,
I don't know when that will be.

-- 
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: 604 bytes --]

Re: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Tue, 17 Jan 2017 20:08:17, Eric Blake wrote:
> [again, your email client likes to break threading, so I nearly missed
> this one]

Sorry about that, I wrote my own client so that I could reply without
subscribing:

http://github.com/svnpenn/a/blob/master/etc/mailing-list.awk

I switched the "In-Reply-To" header to "References", hopefully that will fix it.

> Are you building readline from source each step of the bisection?

Yes.

> are you applying all the downstream cygwin patches each build, or just using
> the stock upstream sources?

The latter.

> Normally, I _don't_ run anything under cmd.exe (all my use of cygwin is
> under mintty), so I need a lot of hand-holding to reproduce anything in
> your setup, in particular how you are even getting extended characters
> to print (what locales and/or window settings you have to set).

1. You must run as a login shell, i.e. "bash -l"

2. You must have in /etc/profile, .profile or similar:

LANG=en_US.UTF-8


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Doug Henderson]

On 17 January 2017 at 20:55, Steven Penny  wrote:
> On Tue, 17 Jan 2017 20:08:17, Eric Blake wrote:
>> [again, your email client likes to break threading, so I nearly missed
>> this one]
>
> Sorry about that, I wrote my own client so that I could reply without
> subscribing:

I find it despicable that you want us to participating in solving a
problem you encountered, but seem unwilling to subscribe so you can
participate in solving problems that the rest of us encounter.

I feel that you are using false pretenses to use our time to think
about your problem and offer solutions, if we can, without
reciprocating.

Shame on you.

Doug

--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Larry Hall (Cygwin)]

On 01/17/2017 11:09 PM, Doug Henderson wrote:
> On 17 January 2017 at 20:55, Steven Penny  wrote:
>> On Tue, 17 Jan 2017 20:08:17, Eric Blake wrote:
>>> [again, your email client likes to break threading, so I nearly missed
>>> this one]
>>
>> Sorry about that, I wrote my own client so that I could reply without
>> subscribing:
>
> I find it despicable that you want us to participating in solving a
> problem you encountered, but seem unwilling to subscribe so you can
> participate in solving problems that the rest of us encounter.
>
> I feel that you are using false pretenses to use our time to think
> about your problem and offer solutions, if we can, without
> reciprocating.
>
> Shame on you.

Whoa!  While I agree there has been some very negatively charged comments
coming from Steven in this thread, I don't think we need to start finding
fault with everything he says and does relative to it.  It's perfectly
fine for him, or anyone else for that matter, to monitor a thread or this
list in whatever form works for them.  In this case, Steven corrected the
problem when it was pointed out to him that his method was breaking
threading.  That's a perfectly acceptable response and all we can ask of
him in that regard.  There's no need to go looking for fault in that.

Let's try to keep this thread productive and moving forward.  While I
would say, in my estimation, it got off to a rocky start, I think the
recent exchange between Steven and Eric has done allot to track the
problem and to scale back the negative comments.  Let's try to keep the
thread moving forward on both fronts.


-- 
Larry

_____________________________________________________________________

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

-- 
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Wed, 18 Jan 2017 00:02:37, "Larry Hall (Cygwin)" wrote:
> Whoa!  While I agree there has been some very negatively charged comments
> coming from Steven in this thread, I don't think we need to start finding
> fault with everything he says and does relative to it.  It's perfectly
> fine for him, or anyone else for that matter, to monitor a thread or this
> list in whatever form works for them.  In this case, Steven corrected the
> problem when it was pointed out to him that his method was breaking
> threading.  That's a perfectly acceptable response and all we can ask of
> him in that regard.  There's no need to go looking for fault in that.

Thanks. I was put off by Eric at first, as it seemed he wanted me to do
everything, and I was not looking forward to fixing a low level bug like this.
Still am not.

> Let's try to keep this thread productive and moving forward.  While I
> would say, in my estimation, it got off to a rocky start, I think the
> recent exchange between Steven and Eric has done allot to track the
> problem and to scale back the negative comments.  Let's try to keep the
> thread moving forward on both fronts.

I would like to keep this productive; my next step is to try to unpack that
large commit, while also helping Eric to reproduce.

On Tue, 17 Jan 2017 21:09:33, Doug Henderson wrote:
> I find it despicable that you want us to participating in solving a
> problem you encountered, but seem unwilling to subscribe so you can
> participate in solving problems that the rest of us encounter.

I am going to keep my response here civil, with repsect to Larry. I would like
you to know that I have answered more Cygwin questions on Stack Overflow than
anyone on the site:

http://stackoverflow.com/tags/cygwin/topusers

No, I am no subscribed to this list, but the point of my script is to allow me
to reply without subscribing. I can reply to any threads, not just my own.


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Tue, 17 Jan 2017 21:45:07, Steven Penny wrote:
> I would like to keep this productive; my next step is to try to unpack that
> large commit, while also helping Eric to reproduce.

I have root caused this. With the new readline, running ./configure creates
config.h, as expected. However new version adds this line:

#define HAVE_PSELECT 1

If you remove this from config.h, then make, the resultant DLL behaves as it did
with previous readline. Not sure what is causing this change, I will try to dig
into it more.


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Eliot Moss]

On 1/19/2017 12:22 AM, Steven Penny wrote:
 > On Tue, 17 Jan 2017 21:45:07, Steven Penny wrote:
 >> I would like to keep this productive; my next step is to try to unpack that
 >> large commit, while also helping Eric to reproduce.
 >
 > I have root caused this. With the new readline, running ./configure creates
 > config.h, as expected. However new version adds this line:
 >
 > #define HAVE_PSELECT 1
 >
 > If you remove this from config.h, then make, the resultant DLL behaves as it did
 > with previous readline. Not sure what is causing this change, I will try to dig
 > into it more.

You probably found it, but here's why they added use of pselect:

"h.  Use pselect(2), if available, to wait for input before calling read(2), so
     a SIGWINCH can interrupt it, since it doesn't interrupt read(2)."

Of course that does not explain why it's not working how we would like ...

Regards - EM

--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Eric Blake]

[-- Attachment #1.1: Type: text/plain, Size: 2423 bytes --]

On 01/19/2017 07:23 AM, Eliot Moss wrote:
>> I have root caused this. With the new readline, running ./configure
> creates
>> config.h, as expected. However new version adds this line:
>>
>> #define HAVE_PSELECT 1
>>
>> If you remove this from config.h, then make, the resultant DLL behaves
> as it did
>> with previous readline. Not sure what is causing this change, I will
> try to dig
>> into it more.
> 
> You probably found it, but here's why they added use of pselect:
> 
> "h.  Use pselect(2), if available, to wait for input before calling
> read(2), so
>     a SIGWINCH can interrupt it, since it doesn't interrupt read(2)."
> 
> Of course that does not explain why it's not working how we would like ...

Here's my guess - cygwin1.dll has an incomplete implementation of
pselect on Windows consoles when it comes to capturing ALT-numkeypad
sequences.  Probably something to do with the fact that it takes
multiple keystrokes before a single character is supposed to be
recognized, which means you have to make a decision after each key state
change event whether a character is complete or not.  Under the old
code, when pselect was not in use, readline just used read() until a
character showed up (but blocked in doing so, and risked SIGWINCH
issues); now readline can use pselect() and sees that a key has been
pressed, except that there is no input available until several more keys
are pressed and a key released.

I'm no expert in the windows console handling code in cygwin1.dll, but
if you are correct that the use (or absence) of HAVE_PSELECT in modern
readline makes a difference, then it sounds like we have two things to
do: first, I will rebuild readline with a patch to avoid pselect() (the
behavior of readline will be riskier, and may break under SIGWINCH, but
at least that is no worse than it was with the old readline), meanwhile,
someone with more expertise in the console handling code will need a
boiled-down testcase (probably a simple .c program that compares a
straight read() with a pselect() loop that decides when to read()) to
see if they can fix the handling of ALT-numkeypad entries under the
console in cygwin1.dll.  Once a fixed cygwin1.dll is available, then I
can re-release a version of readline built with HAVE_PSELECT.

-- 
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: 604 bytes --]

Re: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 2881 bytes --]

On Jan 19 08:21, Eric Blake wrote:
> On 01/19/2017 07:23 AM, Eliot Moss wrote:
> >> I have root caused this. With the new readline, running ./configure
> > creates
> >> config.h, as expected. However new version adds this line:
> >>
> >> #define HAVE_PSELECT 1
> >>
> >> If you remove this from config.h, then make, the resultant DLL behaves
> > as it did
> >> with previous readline. Not sure what is causing this change, I will
> > try to dig
> >> into it more.
> > 
> > You probably found it, but here's why they added use of pselect:
> > 
> > "h.  Use pselect(2), if available, to wait for input before calling
> > read(2), so
> >     a SIGWINCH can interrupt it, since it doesn't interrupt read(2)."
> > 
> > Of course that does not explain why it's not working how we would like ...
> 
> Here's my guess - cygwin1.dll has an incomplete implementation of
> pselect on Windows consoles when it comes to capturing ALT-numkeypad
> sequences.  Probably something to do with the fact that it takes
> multiple keystrokes before a single character is supposed to be
> recognized, which means you have to make a decision after each key state
> change event whether a character is complete or not.  Under the old
> code, when pselect was not in use, readline just used read() until a
> character showed up (but blocked in doing so, and risked SIGWINCH
> issues); now readline can use pselect() and sees that a key has been
> pressed, except that there is no input available until several more keys
> are pressed and a key released.
> 
> I'm no expert in the windows console handling code in cygwin1.dll, but
> if you are correct that the use (or absence) of HAVE_PSELECT in modern
> readline makes a difference, then it sounds like we have two things to
> do: first, I will rebuild readline with a patch to avoid pselect() (the
> behavior of readline will be riskier, and may break under SIGWINCH, but
> at least that is no worse than it was with the old readline), meanwhile,
> someone with more expertise in the console handling code will need a
> boiled-down testcase (probably a simple .c program that compares a
> straight read() with a pselect() loop that decides when to read()) to
> see if they can fix the handling of ALT-numkeypad entries under the
> console in cygwin1.dll.  Once a fixed cygwin1.dll is available, then I
> can re-release a version of readline built with HAVE_PSELECT.

I applied a patch to Cygwin to handle these Alt-Numpad sequences, see
https://sourceware.org/git/?p=newlib-cygwin.git;a=commitdiff;h=4652cc4

I uploaded new developer snapshots to https://cygwin.com/snapshots/
for testing.  Please give them a try and report back.


Thanks,
Corinna

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

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Thu, 19 Jan 2017 19:13:35, Corinna Vinschen wrote:
> I applied a patch to Cygwin to handle these Alt-Numpad sequences, see
> https://sourceware.org/git/?p=3Dnewlib-cygwin.git;a=3Dcommitdiff;h=3D4652cc4
> I uploaded new developer snapshots to https://cygwin.com/snapshots/
> for testing.  Please give them a try and report back.

This fixes it for me. I should also mention that my fix mentioned earlier, and
now this patch, also fix pasting as well. Thanks.


--
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: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 709 bytes --]

On Jan 19 15:53, Steven Penny wrote:
> On Thu, 19 Jan 2017 19:13:35, Corinna Vinschen wrote:
> > I applied a patch to Cygwin to handle these Alt-Numpad sequences, see
> > https://sourceware.org/git/?p=3Dnewlib-cygwin.git;a=3Dcommitdiff;h=3D4652cc4
> > I uploaded new developer snapshots to https://cygwin.com/snapshots/
> > for testing.  Please give them a try and report back.
> 
> This fixes it for me. I should also mention that my fix mentioned earlier, and
> now this patch, also fix pasting as well. Thanks.

Thanks for testing!


Corinna

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

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Eric Blake]

[-- Attachment #1.1: Type: text/plain, Size: 1094 bytes --]

On 01/18/2017 11:22 PM, Steven Penny wrote:
> On Tue, 17 Jan 2017 21:45:07, Steven Penny wrote:
>> I would like to keep this productive; my next step is to try to unpack that
>> large commit, while also helping Eric to reproduce.
> 
> I have root caused this. With the new readline, running ./configure creates
> config.h, as expected. However new version adds this line:
> 
> #define HAVE_PSELECT 1
> 
> If you remove this from config.h, then make, the resultant DLL behaves as it did
> with previous readline. Not sure what is causing this change, I will try to dig
> into it more.

I've uploaded a test release of readline-7.0.1-2 which disables pselect
support. I would appreciate if you could test that it fixes things for
you without using the official release cygwin1.dll (although you have
already tested that the recent test build of cygwin also fixes things).

Once we have another cygwin release, I'll build readline-7.0.1-3 which
re-enables pselect.

-- 
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: 604 bytes --]

Re: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Eric Blake]

[-- Attachment #1.1: Type: text/plain, Size: 1491 bytes --]

On 01/20/2017 08:51 AM, Eric Blake wrote:
> On 01/18/2017 11:22 PM, Steven Penny wrote:
>> On Tue, 17 Jan 2017 21:45:07, Steven Penny wrote:
>>> I would like to keep this productive; my next step is to try to unpack that
>>> large commit, while also helping Eric to reproduce.
>>
>> I have root caused this. With the new readline, running ./configure creates
>> config.h, as expected. However new version adds this line:
>>
>> #define HAVE_PSELECT 1
>>
>> If you remove this from config.h, then make, the resultant DLL behaves as it did
>> with previous readline. Not sure what is causing this change, I will try to dig
>> into it more.
> 
> I've uploaded a test release of readline-7.0.1-2 which disables pselect
> support. I would appreciate if you could test that it fixes things for
> you without using the official release cygwin1.dll (although you have

s/without//

> already tested that the recent test build of cygwin also fixes things).

(that is, please test that bash-4.4.5-1, cygwin-2.6.1-1, and
libreadline7-7.0.1-2 should work; you've already tested that
bash-4.4.5-1, libreadline7-7.0.1-1, and cygwin1-20170119 snapshot
worked; and likewise mixing libreadline7-7.0.1-2 and the snapshot should
work although it is less important)

> 
> Once we have another cygwin release, I'll build readline-7.0.1-3 which
> re-enables pselect.
> 

-- 
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: 604 bytes --]

Re: [ANNOUNCEMENT] Updated: libreadline7-7.0.1-1, libreadline-devel-7.0.1-1, bash-4.4.5-1

[Steven Penny]

On Fri, 20 Jan 2017 09:00:10, Eric Blake wrote:
> (that is, please test that bash-4.4.5-1, cygwin-2.6.1-1, and
> libreadline7-7.0.1-2 should work; you've already tested that
> bash-4.4.5-1, libreadline7-7.0.1-1, and cygwin1-20170119 snapshot
> worked; and likewise mixing libreadline7-7.0.1-2 and the snapshot should
> work although it is less important)

TEST 1:

    $ cygcheck -sv | awk '$1~/^(bash|cygwin|libreadline7|"cygwin1.dll")$/'
                      "cygwin1.dll" v0.0 ts=2016-12-16 04:55
    bash                    4.4.5-1            OK
    cygwin                  2.6.1-1            OK
    libreadline7            7.0.1-2            OK

RESULT:

    Alt CP437 OK
    Alt UTF-8 OK
    Pasting OK

TEST 2:

    $ cygcheck -sv | awk '$1~/^(bash|cygwin|libreadline7|"cygwin1.dll")$/'
                      "cygwin1.dll" v0.0 ts=2017-01-19 15:16
    bash                    4.4.5-1            OK
    cygwin                  2.6.1-1            OK
    libreadline7            7.0.1-2            OK

RESULT:

    Alt CP437 OK
    Alt UTF-8 OK
    Pasting OK


--
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