[ANNOUNCEMENT] Test: cygutils 1.4.16-4 (TEST)

[ANNOUNCEMENT] Test: cygutils 1.4.16-4 (TEST)

[Mark Geisert]

The following test packages have been uploaded to the Cygwin distribution:

* cygutils-1.4.16-4
* cygutils-extra-1.4.16-4
* cygutils-x11-1.4.16-4

This release fixes a couple of issues reported in -3 and re-implements
getclip's capture of Unicode characters from the Windows clipboard using
a different method.  This was required because the original approach
would not work in 32-bit Cygwin operating on 64-bit Windows.

IF YOU USE GETCLIP TO CAPTURE NON-ASCII CHARACTERS FROM THE WINDOWS
CLIPBOARD, PLEASE TEST THIS NEWEST GETCLIP TO MAKE SURE IT WORKS FOR YOU.
Report any issues to the main Cygwin mailing list.

Excerpt from ChangeLog follows...
2021-08-15  Mark Geisert  <mark@maxrnd.com>  (1.4.16-4 release)

    * Makefile.am: Add -lkernel32 to src_clip_getclip_LDADD.  This somehow
    didn't make it into the released -3 package.  Thank you Takashi Yano.
    * Fix maintainer build environment issues that caused cygstart crash
    on non-AVX user systems: Thank you Takashi Yano, Brian Inglis, Achim
    Gratz and user Jay Abel for reporting the issue.
    * src/cygdrop/cygdrop.cc: Fix crash with recent gcc and fix printf
    format strings. Thank you Christian Franke.

2021-07-05  Mark Geisert  <mark@maxrnd.com>  (1.4.16-3 release)

    * src/clip/getclip.c: Add Unicode support.

In about two weeks time this package will be marked current unless bugs
are found during this test period.
More Happy Clipping,

..mark

Re: [ANNOUNCEMENT] Test: cygutils 1.4.16-4 (TEST)

[Takashi Yano]

On Tue, 17 Aug 2021 00:11:51 -0700
Mark Geisert wrote:
> The following test packages have been uploaded to the Cygwin distribution:
> 
> * cygutils-1.4.16-4
> * cygutils-extra-1.4.16-4
> * cygutils-x11-1.4.16-4
> 
> This release fixes a couple of issues reported in -3 and re-implements
> getclip's capture of Unicode characters from the Windows clipboard using
> a different method.  This was required because the original approach
> would not work in 32-bit Cygwin operating on 64-bit Windows.
> 
> IF YOU USE GETCLIP TO CAPTURE NON-ASCII CHARACTERS FROM THE WINDOWS
> CLIPBOARD, PLEASE TEST THIS NEWEST GETCLIP TO MAKE SURE IT WORKS FOR YOU.
> Report any issues to the main Cygwin mailing list.
> 
> Excerpt from ChangeLog follows...
> 2021-08-15  Mark Geisert  <mark@maxrnd.com>  (1.4.16-4 release)
> 
>     * Makefile.am: Add -lkernel32 to src_clip_getclip_LDADD.  This somehow
>     didn't make it into the released -3 package.  Thank you Takashi Yano.
>     * Fix maintainer build environment issues that caused cygstart crash
>     on non-AVX user systems: Thank you Takashi Yano, Brian Inglis, Achim
>     Gratz and user Jay Abel for reporting the issue.
>     * src/cygdrop/cygdrop.cc: Fix crash with recent gcc and fix printf
>     format strings. Thank you Christian Franke.
> 
> 2021-07-05  Mark Geisert  <mark@maxrnd.com>  (1.4.16-3 release)
> 
>     * src/clip/getclip.c: Add Unicode support.
> 
> In about two weeks time this package will be marked current unless bugs
> are found during this test period.
> More Happy Clipping,

Sorry for late response.

I tested the 1.4.16-4(TEST), and confirmed that cygstart, getclip
and putclip can be executed on my old machine which does not have
AVX instruction.

However, I encountered a problem with putclip.
Run 'echo AAAAAAAA | putclip' and paste clipboard to notepad.exe.
Pasted string is garbled in my environment.

What am I wrong with?

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: [ANNOUNCEMENT] Test: cygutils 1.4.16-4 (TEST)

[Mark Geisert]

Hi Takashi,

Takashi Yano via Cygwin wrote:
> On Tue, 17 Aug 2021 00:11:51 -0700
> Mark Geisert wrote:
>> The following test packages have been uploaded to the Cygwin distribution:
>>
>> * cygutils-1.4.16-4
>> * cygutils-extra-1.4.16-4
>> * cygutils-x11-1.4.16-4
>>
>> This release fixes a couple of issues reported in -3 and re-implements
>> getclip's capture of Unicode characters from the Windows clipboard using
>> a different method.  This was required because the original approach
>> would not work in 32-bit Cygwin operating on 64-bit Windows.
>>
>> IF YOU USE GETCLIP TO CAPTURE NON-ASCII CHARACTERS FROM THE WINDOWS
>> CLIPBOARD, PLEASE TEST THIS NEWEST GETCLIP TO MAKE SURE IT WORKS FOR YOU.
>> Report any issues to the main Cygwin mailing list.
[...]
>> In about two weeks time this package will be marked current unless bugs
>> are found during this test period.
[...]
> I tested the 1.4.16-4(TEST), and confirmed that cygstart, getclip
> and putclip can be executed on my old machine which does not have
> AVX instruction.

Thanks for testing!

> However, I encountered a problem with putclip.
> Run 'echo AAAAAAAA | putclip' and paste clipboard to notepad.exe.
> Pasted string is garbled in my environment.
> 
> What am I wrong with?

s/am I/are we/ :-\  I can reproduce this.  No idea what's wrong as there were no 
recent changes to putclip.  Let me dig further.  Is this something you've done 
routinely in the past and it was known to work?
Thanks for the report,

..mark

Re: [ANNOUNCEMENT] Test: cygutils 1.4.16-4 (TEST)

[Takashi Yano]

On Fri, 20 Aug 2021 16:28:42 -0700
Mark Geisert wrote:
> Hi Takashi,
> 
> Takashi Yano via Cygwin wrote:
> > On Tue, 17 Aug 2021 00:11:51 -0700
> > Mark Geisert wrote:
> >> The following test packages have been uploaded to the Cygwin distribution:
> >>
> >> * cygutils-1.4.16-4
> >> * cygutils-extra-1.4.16-4
> >> * cygutils-x11-1.4.16-4
> >>
> >> This release fixes a couple of issues reported in -3 and re-implements
> >> getclip's capture of Unicode characters from the Windows clipboard using
> >> a different method.  This was required because the original approach
> >> would not work in 32-bit Cygwin operating on 64-bit Windows.
> >>
> >> IF YOU USE GETCLIP TO CAPTURE NON-ASCII CHARACTERS FROM THE WINDOWS
> >> CLIPBOARD, PLEASE TEST THIS NEWEST GETCLIP TO MAKE SURE IT WORKS FOR YOU.
> >> Report any issues to the main Cygwin mailing list.
> [...]
> >> In about two weeks time this package will be marked current unless bugs
> >> are found during this test period.
> [...]
> > I tested the 1.4.16-4(TEST), and confirmed that cygstart, getclip
> > and putclip can be executed on my old machine which does not have
> > AVX instruction.
> 
> Thanks for testing!
> 
> > However, I encountered a problem with putclip.
> > Run 'echo AAAAAAAA | putclip' and paste clipboard to notepad.exe.
> > Pasted string is garbled in my environment.
> > 
> > What am I wrong with?
> 
> s/am I/are we/ :-\  I can reproduce this.  No idea what's wrong as there were no 
> recent changes to putclip.  Let me dig further.  Is this something you've done 
> routinely in the past and it was known to work?

No. But I confirmed that this issue does not happen with 1.4.16-2.
1.4.16-3 and 1.4.16-4(TEST) have this problem.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: [ANNOUNCEMENT] Test: cygutils 1.4.16-4 (TEST)

[Christian Franke]

Takashi Yano via Cygwin wrote:
> On Fri, 20 Aug 2021 16:28:42 -0700
> Mark Geisert wrote:
>> Hi Takashi,
>>
>> Takashi Yano via Cygwin wrote:
>> ...
>>> However, I encountered a problem with putclip.
>>> Run 'echo AAAAAAAA | putclip' and paste clipboard to notepad.exe.
>>> Pasted string is garbled in my environment.
>>>
>>> What am I wrong with?
>> s/am I/are we/ :-\  I can reproduce this.  No idea what's wrong as there were no
>> recent changes to putclip.  Let me dig further.  Is this something you've done
>> routinely in the past and it was known to work?
> No. But I confirmed that this issue does not happen with 1.4.16-2.
> 1.4.16-3 and 1.4.16-4(TEST) have this problem.

The problem is the putclip addition in file 
"cygutils-1.4.16-4.src.patch" from the src package.

It adds "SetClipboardData (CF_UNICODETEXT, hData)" but the data is not 
in 16 bit unicode format. Removing this new section fixes the problem.

Another unrelated issue: GlobalFree() should possibly not be called on 
handles passed to clipboard. See comment in "fhandler_clipboard.cc" from 
Cygwin source.

BTW: The cygdrop crash does no longer occur.

BTW^2: Is there a new public repo for cygutils?
The old one ends at 1.4.16-2:
https://sourceware.org/git/?p=cygwin-apps/cygutils.git;a=summary

Regards,
Christian

Re: [ANNOUNCEMENT] Test: cygutils 1.4.16-4 (TEST)

[Corinna Vinschen]

On Aug 23 19:48, Christian Franke wrote:
> Takashi Yano via Cygwin wrote:
> > On Fri, 20 Aug 2021 16:28:42 -0700
> > Mark Geisert wrote:
> > > Hi Takashi,
> > > 
> > > Takashi Yano via Cygwin wrote:
> > > ...
> > > > However, I encountered a problem with putclip.
> > > > Run 'echo AAAAAAAA | putclip' and paste clipboard to notepad.exe.
> > > > Pasted string is garbled in my environment.
> > > > 
> > > > What am I wrong with?
> > > s/am I/are we/ :-\  I can reproduce this.  No idea what's wrong as there were no
> > > recent changes to putclip.  Let me dig further.  Is this something you've done
> > > routinely in the past and it was known to work?
> > No. But I confirmed that this issue does not happen with 1.4.16-2.
> > 1.4.16-3 and 1.4.16-4(TEST) have this problem.
> 
> The problem is the putclip addition in file "cygutils-1.4.16-4.src.patch"
> from the src package.
> 
> It adds "SetClipboardData (CF_UNICODETEXT, hData)" but the data is not in 16
> bit unicode format. Removing this new section fixes the problem.
> 
> Another unrelated issue: GlobalFree() should possibly not be called on
> handles passed to clipboard. See comment in "fhandler_clipboard.cc" from
> Cygwin source.
> 
> BTW: The cygdrop crash does no longer occur.
> 
> BTW^2: Is there a new public repo for cygutils?
> The old one ends at 1.4.16-2:
> https://sourceware.org/git/?p=cygwin-apps/cygutils.git;a=summary

Theoretically the idea was that this is the official repo for
this tool.  Mark, please keep it up.


Thanks,
Corinna

Re: [ANNOUNCEMENT] Test: cygutils 1.4.16-4 (TEST)

[Mark Geisert]

Hi Christian,

Christian Franke wrote:
> Takashi Yano via Cygwin wrote:
>> On Fri, 20 Aug 2021 16:28:42 -0700
>> Mark Geisert wrote:
>>> Hi Takashi,
>>>
>>> Takashi Yano via Cygwin wrote:
>>> ...
>>>> However, I encountered a problem with putclip.
>>>> Run 'echo AAAAAAAA | putclip' and paste clipboard to notepad.exe.
>>>> Pasted string is garbled in my environment.
>>>>
>>>> What am I wrong with?
>>> s/am I/are we/ :-\  I can reproduce this.  No idea what's wrong as there were no
>>> recent changes to putclip.  Let me dig further.  Is this something you've done
>>> routinely in the past and it was known to work?
>> No. But I confirmed that this issue does not happen with 1.4.16-2.
>> 1.4.16-3 and 1.4.16-4(TEST) have this problem.
> 
> The problem is the putclip addition in file "cygutils-1.4.16-4.src.patch" from the 
> src package.
> 
> It adds "SetClipboardData (CF_UNICODETEXT, hData)" but the data is not in 16 bit 
> unicode format. Removing this new section fixes the problem.

Alternatively, add the missing code to make it work.  I'm debugging that now.

> Another unrelated issue: GlobalFree() should possibly not be called on handles 
> passed to clipboard. See comment in "fhandler_clipboard.cc" from Cygwin source.

I appreciate the help.

> BTW: The cygdrop crash does no longer occur.

Right, a separate issue with my build environment that has been fixed.

> BTW^2: Is there a new public repo for cygutils?
> The old one ends at 1.4.16-2:
> https://sourceware.org/git/?p=cygwin-apps/cygutils.git;a=summary

The public repo will be updated when I've got something stable and correct enough 
to place there ;-).
Thanks & Regards,

..mark

Re: [ANNOUNCEMENT] Test: cygutils 1.4.16-4 (TEST)

[Mark Geisert]

Hi Corinna,

Corinna Vinschen via Cygwin wrote:
> On Aug 23 19:48, Christian Franke wrote:
>> Takashi Yano via Cygwin wrote:
>>> On Fri, 20 Aug 2021 16:28:42 -0700
>>> Mark Geisert wrote:
>>>> Hi Takashi,
>>>>
>>>> Takashi Yano via Cygwin wrote:
>>>> ...
>>>>> However, I encountered a problem with putclip.
>>>>> Run 'echo AAAAAAAA | putclip' and paste clipboard to notepad.exe.
>>>>> Pasted string is garbled in my environment.
>>>>>
>>>>> What am I wrong with?
>>>> s/am I/are we/ :-\  I can reproduce this.  No idea what's wrong as there were no
>>>> recent changes to putclip.  Let me dig further.  Is this something you've done
>>>> routinely in the past and it was known to work?
>>> No. But I confirmed that this issue does not happen with 1.4.16-2.
>>> 1.4.16-3 and 1.4.16-4(TEST) have this problem.
>>
>> The problem is the putclip addition in file "cygutils-1.4.16-4.src.patch"
>> from the src package.
>>
>> It adds "SetClipboardData (CF_UNICODETEXT, hData)" but the data is not in 16
>> bit unicode format. Removing this new section fixes the problem.
>>
>> Another unrelated issue: GlobalFree() should possibly not be called on
>> handles passed to clipboard. See comment in "fhandler_clipboard.cc" from
>> Cygwin source.
>>
>> BTW: The cygdrop crash does no longer occur.
>>
>> BTW^2: Is there a new public repo for cygutils?
>> The old one ends at 1.4.16-2:
>> https://sourceware.org/git/?p=cygwin-apps/cygutils.git;a=summary
> 
> Theoretically the idea was that this is the official repo for
> this tool.  Mark, please keep it up.

Yes, that's the intent.  I was tripped up by my build environment not being stock; 
that's been remedied.  I was also mightily confused by Charles' apparent use of 
git *and* a huge patch file.  Some of my previous updates went one place rather 
than both.  Or as in this latest putclip case, into the patch file when it wasn't 
yet complete.  I am in the process of clearing the debris and making it right.
Thanks & Regards,

..mark

Re: [ANNOUNCEMENT] Test: cygutils 1.4.16-4 (TEST)

[Mark Geisert]

Mark Geisert clarifies what he wrote:
> I was also mightily confused by Charles' apparent use of 
> git *and* a huge patch file.

That's not fair.  Charles used cvs; his cygutils repository was converted to git 
and I've been maintaining it since the conversion.  My confusion is in whether 
updates are totally described by git commits or whether they also/instead appear 
in the huge patch file.  This is an internal issue that I need to resolve.

I apologize that problematic cygutils builds have been released due to my not 
fully understanding the subtleties of Charles' build factory.

There's an upcoming cygutils-1.4.16-5 that's intended to be a clean update that 
fixes the reported problems in -3 and -4test.  Once that is out, the public 
repository for cygutils will be updated in short order.
Thanks,

..mark

Re: [ANNOUNCEMENT] Test: cygutils 1.4.16-4 (TEST)

[Corinna Vinschen]

On Aug 23 23:52, Mark Geisert wrote:
> Mark Geisert clarifies what he wrote:
> > I was also mightily confused by Charles' apparent use of git *and* a
> > huge patch file.
> 
> That's not fair.  Charles used cvs; his cygutils repository was converted to
> git and I've been maintaining it since the conversion.  My confusion is in
> whether updates are totally described by git commits or whether they
> also/instead appear in the huge patch file.  This is an internal issue that
> I need to resolve.

Ideally the public repo follows the development closely, including
bugs and mistakes.  We all learn only from those :)  IMHO it could
be pretty helpful to feed the changes from the patch file in small
chunks into the repo, and then your follow up patches and reversions
afterwards.  So a history stays visible, even if partially faked.

In case of cygutils, given how this is a Cygwin-only tool and we have
full control over it, the public repo should be the one-stop-shop for
code.  There's no reason for a non-git patch file at all these days.

> I apologize that problematic cygutils builds have been released due to my
> not fully understanding the subtleties of Charles' build factory.

No reason to apologize!  You wrote in your mail how Chuck had his stuff
in CVS originally.  Additionally the stuff wasn't packed with cygport
at one point.  All this had a big influence on how the files were
packed.  I, for one, am grateful that you pull the package into modern
times.

> There's an upcoming cygutils-1.4.16-5 that's intended to be a clean update
> that fixes the reported problems in -3 and -4test.  Once that is out, the
> public repository for cygutils will be updated in short order.
> Thanks,

Thanks to you,
Corinna