Bug with Cygwin's 'quilt'

Bug with Cygwin's 'quilt'

[Matt D.]

The last e-mail I supplied to the mailing list was missing the link to 
the example. See this one for the link.

---

There seems to be a problem with Cygwin's quilt. This simple example 
simply alters a #define from 'MESAGL' to 'NX_MESAGL'. That's it.

New quilt created, ok.
New patch, ok.
Edit file, ok.
Refresh (create patch), ok.
Rollback changes, ok.
Reapply patch.. error:

   >>> quilt push
   >>> Applying patch test.patch
   >>> patching file Imakefile
   >>> Hunk #1 FAILED at 2.
   >>> 1 out of 1 hunk FAILED -- rejects in file Imakefile
   >>> Patch test.patch does not apply (enforce with -f)

http://codespunk.com/files/upload/cygwin_quilt_00.zip

Extract to a directory, cd in, and run 'quilt push' to generate the error.



--
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: Bug with Cygwin's 'quilt'

[Matt D.]

Built the latest source 0.60 (same version as Cygwin) from 
http://freecode.com/projects/quilt. Built on CentOS 6.4 and passes my 
previously provided test just fine.

Downloaded the same source to Cygwin, rebuilt, replaced quilt in /bin; 
test still errors out. I also tried the latest cygwin1.dll snapshot; 
same problem.


On 6/18/2013 1:09 AM, Matt D. wrote:
> The last e-mail I supplied to the mailing list was missing the link to
> the example. See this one for the link.
>
> ---
>
> There seems to be a problem with Cygwin's quilt. This simple example
> simply alters a #define from 'MESAGL' to 'NX_MESAGL'. That's it.
>
> New quilt created, ok.
> New patch, ok.
> Edit file, ok.
> Refresh (create patch), ok.
> Rollback changes, ok.
> Reapply patch.. error:
>
>  >>> quilt push
>  >>> Applying patch test.patch
>  >>> patching file Imakefile
>  >>> Hunk #1 FAILED at 2.
>  >>> 1 out of 1 hunk FAILED -- rejects in file Imakefile
>  >>> Patch test.patch does not apply (enforce with -f)
>
> http://codespunk.com/files/upload/cygwin_quilt_00.zip
>
> Extract to a directory, cd in, and run 'quilt push' to generate the error.
>
>
>
> --
> 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: Bug with Cygwin's 'quilt' is actually in 'patch'

[Matt D.]

I've been looking further into this and it appears as though the problem 
is in 'patch' not 'quilt'. quilt is actually a collection of bash 
scripts and calls patch to do the actual patching.

Using the same example I provided earlier in the thread, the same error 
occurs when calling patch directly:

$ patch Imakefile patches/test.patch

Running dos2unix on test.patch will allow the patch to apply 
successfully. However, this is WRONG. Imakefile and the initially 
created test.patch both use CRLF line endings. The patch should 
definitely NOT apply by introducing actual disparity.

To summarize, the patch to Imakefile (CRLF) will apply if it is 
converted to LF line endings. Using the '--binary' switch seems to be a 
workaround for this issue.


On 6/18/2013 1:36 AM, Matt D. wrote:
> Built the latest source 0.60 (same version as Cygwin) from
> http://freecode.com/projects/quilt. Built on CentOS 6.4 and passes my
> previously provided test just fine.
>
> Downloaded the same source to Cygwin, rebuilt, replaced quilt in /bin;
> test still errors out. I also tried the latest cygwin1.dll snapshot;
> same problem.
>
>
> On 6/18/2013 1:09 AM, Matt D. wrote:
>> The last e-mail I supplied to the mailing list was missing the link to
>> the example. See this one for the link.
>>
>> ---
>>
>> There seems to be a problem with Cygwin's quilt. This simple example
>> simply alters a #define from 'MESAGL' to 'NX_MESAGL'. That's it.
>>
>> New quilt created, ok.
>> New patch, ok.
>> Edit file, ok.
>> Refresh (create patch), ok.
>> Rollback changes, ok.
>> Reapply patch.. error:
>>
>> >>> quilt push
>> >>> Applying patch test.patch
>> >>> patching file Imakefile
>> >>> Hunk #1 FAILED at 2.
>> >>> 1 out of 1 hunk FAILED -- rejects in file Imakefile
>> >>> Patch test.patch does not apply (enforce with -f)
>>
>> http://codespunk.com/files/upload/cygwin_quilt_00.zip
>>
>> Extract to a directory, cd in, and run 'quilt push' to generate the
>> error.
>>
>>
>>
>> --
>> 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
>
>
>

--
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: Bug with Cygwin's 'quilt' is actually in 'patch'

[Christopher Faylor]

On Wed, Jun 19, 2013 at 11:31:48PM -0400, Matt D. wrote:
>I've been looking further into this and it appears as though the problem 
>is in 'patch' not 'quilt'. quilt is actually a collection of bash 
>scripts and calls patch to do the actual patching.
>
>Using the same example I provided earlier in the thread, the same error 
>occurs when calling patch directly:
>
>$ patch Imakefile patches/test.patch
>
>Running dos2unix on test.patch will allow the patch to apply 
>successfully. However, this is WRONG. Imakefile and the initially 
>created test.patch both use CRLF line endings. The patch should 
>definitely NOT apply by introducing actual disparity.
>
>To summarize, the patch to Imakefile (CRLF) will apply if it is 
>converted to LF line endings. Using the '--binary' switch seems to be a 
>workaround for this issue.

Sorry but we're emulating Linux here.  You shouldn't have CRLF endings
on your text file if you want the tools to work reliably.

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: Bug with Cygwin's 'quilt' is actually in 'patch'

[Matt D.]

I'm building from Linux source from the X2Go git repository. The patches 
are being applied downstream to the last base nx libraries provided by 
NoMachine. It can't be helped if the original source has CRLF in this case.

I understand that Cygwin is trying to emulate Linux here, but I don't 
believe that is the appropriate response regarding tools like 'patch' 
which should not have this kind of limitation. The fact that it thinks:

 > \r\n <> \r\n

but..

 > \r\n == \n

As I mentioned previously, patch does NOT have this issue on Linux using 
the EXACT SAME test case.

This is definitely a bug.


On 6/20/2013 1:47 AM, Christopher Faylor wrote:
> On Wed, Jun 19, 2013 at 11:31:48PM -0400, Matt D. wrote:
>> I've been looking further into this and it appears as though the problem
>> is in 'patch' not 'quilt'. quilt is actually a collection of bash
>> scripts and calls patch to do the actual patching.
>>
>> Using the same example I provided earlier in the thread, the same error
>> occurs when calling patch directly:
>>
>> $ patch Imakefile patches/test.patch
>>
>> Running dos2unix on test.patch will allow the patch to apply
>> successfully. However, this is WRONG. Imakefile and the initially
>> created test.patch both use CRLF line endings. The patch should
>> definitely NOT apply by introducing actual disparity.
>>
>> To summarize, the patch to Imakefile (CRLF) will apply if it is
>> converted to LF line endings. Using the '--binary' switch seems to be a
>> workaround for this issue.
>
> Sorry but we're emulating Linux here.  You shouldn't have CRLF endings
> on your text file if you want the tools to work reliably.
>
> 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: Bug with Cygwin's 'quilt' is actually in 'patch'

[Corinna Vinschen]

On Jun 19 23:31, Matt D. wrote:
> I've been looking further into this and it appears as though the
> problem is in 'patch' not 'quilt'. quilt is actually a collection of
> bash scripts and calls patch to do the actual patching.
> 
> Using the same example I provided earlier in the thread, the same
> error occurs when calling patch directly:
> 
> $ patch Imakefile patches/test.patch
> 
> Running dos2unix on test.patch will allow the patch to apply
> successfully. However, this is WRONG. Imakefile and the initially
> created test.patch both use CRLF line endings. The patch should
> definitely NOT apply by introducing actual disparity.
> 
> To summarize, the patch to Imakefile (CRLF) will apply if it is
> converted to LF line endings. Using the '--binary' switch seems to
> be a workaround for this issue.

I can reproduce this problem on 32 bit Cygwin but not on 64 bit Cygwin.

The 64 bit version has a newer patch version 2.7.1, while I so far
neglected to update the 32 bit version which is still on 2.6.1.  I'll
build a new patch 2.7.1 for 32 bit today.  I hope that fixes it for
32 bit as well.  Stay tuned for the announcement.


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 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: Bug with Cygwin's 'quilt' is actually in 'patch'

[Corinna Vinschen]

On Jun 20 09:46, Corinna Vinschen wrote:
> On Jun 19 23:31, Matt D. wrote:
> > I've been looking further into this and it appears as though the
> > problem is in 'patch' not 'quilt'. quilt is actually a collection of
> > bash scripts and calls patch to do the actual patching.
> > 
> > Using the same example I provided earlier in the thread, the same
> > error occurs when calling patch directly:
> > 
> > $ patch Imakefile patches/test.patch
> > 
> > Running dos2unix on test.patch will allow the patch to apply
> > successfully. However, this is WRONG. Imakefile and the initially
> > created test.patch both use CRLF line endings. The patch should
> > definitely NOT apply by introducing actual disparity.
> > 
> > To summarize, the patch to Imakefile (CRLF) will apply if it is
> > converted to LF line endings. Using the '--binary' switch seems to
> > be a workaround for this issue.
> 
> I can reproduce this problem on 32 bit Cygwin but not on 64 bit Cygwin.
> 
> The 64 bit version has a newer patch version 2.7.1, while I so far
> neglected to update the 32 bit version which is still on 2.6.1.  I'll
> build a new patch 2.7.1 for 32 bit today.  I hope that fixes it for
> 32 bit as well.  Stay tuned for the announcement.

Matt?  Ping?  Does the new patch 2.7.1 help?


Thanks,
Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 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: Bug with Cygwin's 'quilt' is actually in 'patch'

[Matt D.]

Thank you. I took the time to make certain that I couldn't find any 
other lingering problems.

The update to 2.7.1 corrects the test case I submitted previously.

Thanks! :)


On 6/21/2013 8:11 AM, Corinna Vinschen wrote:
> On Jun 20 09:46, Corinna Vinschen wrote:
>> On Jun 19 23:31, Matt D. wrote:
>>> I've been looking further into this and it appears as though the
>>> problem is in 'patch' not 'quilt'. quilt is actually a collection of
>>> bash scripts and calls patch to do the actual patching.
>>>
>>> Using the same example I provided earlier in the thread, the same
>>> error occurs when calling patch directly:
>>>
>>> $ patch Imakefile patches/test.patch
>>>
>>> Running dos2unix on test.patch will allow the patch to apply
>>> successfully. However, this is WRONG. Imakefile and the initially
>>> created test.patch both use CRLF line endings. The patch should
>>> definitely NOT apply by introducing actual disparity.
>>>
>>> To summarize, the patch to Imakefile (CRLF) will apply if it is
>>> converted to LF line endings. Using the '--binary' switch seems to
>>> be a workaround for this issue.
>>
>> I can reproduce this problem on 32 bit Cygwin but not on 64 bit Cygwin.
>>
>> The 64 bit version has a newer patch version 2.7.1, while I so far
>> neglected to update the 32 bit version which is still on 2.6.1.  I'll
>> build a new patch 2.7.1 for 32 bit today.  I hope that fixes it for
>> 32 bit as well.  Stay tuned for the announcement.
>
> Matt?  Ping?  Does the new patch 2.7.1 help?
>
>
> Thanks,
> Corinna
>

--
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: Bug with Cygwin's 'quilt' is actually in 'patch'

[Peter B.]

On Jun 20 09:46, Corinna Vinschen wrote:
> I'll build a new patch 2.7.1 for 32 bit today. I hope that fixes it
> for 32 bit as well.

The new 2.7.1 (32bit) brakes CRLF handling for me completely. The old
2.6.1 works properly. 2.7.1 with text files converted from CRLF to LF
works properly as well.

Can you take a look at the download link below? It includes an example
file (with CRLF), the corresponding patch (LF) and the two binaries
(patch v2.6.1 and v2.7.1).

http://www60.zippyshare.com/v/13352963/file.html

# Does work with CRLF
patch261 -Nls Activity.smali < framework.patch

# Does not work with CRLF
patch271 -Nls Activity.smali < framework.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: Bug with Cygwin's 'quilt' is actually in 'patch'

[Corinna Vinschen]

On Jul  3 05:03, Peter B. wrote:
> On Jun 20 09:46, Corinna Vinschen wrote:
> > I'll build a new patch 2.7.1 for 32 bit today. I hope that fixes it
> > for 32 bit as well.
> 
> The new 2.7.1 (32bit) brakes CRLF handling for me completely. The old
> 2.6.1 works properly. 2.7.1 with text files converted from CRLF to LF
> works properly as well.
> 
> Can you take a look at the download link below? It includes an example
> file (with CRLF), the corresponding patch (LF) and the two binaries
> (patch v2.6.1 and v2.7.1).
> 
> http://www60.zippyshare.com/v/13352963/file.html

Please provide a direct link which doesn't require to enable scripting
in a web browser to download.


Thanks,
Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 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: Bug with Cygwin's 'quilt' is actually in 'patch'

[Peter B.]

On 03.07.2013 14:33 Corinna Vinschen wrote:
> On Jul  3 05:03, Peter B. wrote:
>> On Jun 20 09:46, Corinna Vinschen wrote:
>>> I'll build a new patch 2.7.1 for 32 bit today. I hope that fixes it
>>> for 32 bit as well.
>> The new 2.7.1 (32bit) brakes CRLF handling for me completely. The old
>> 2.6.1 works properly. 2.7.1 with text files converted from CRLF to LF
>> works properly as well.
>>
>> Can you take a look at the download link below? It includes an example
>> file (with CRLF), the corresponding patch (LF) and the two binaries
>> (patch v2.6.1 and v2.7.1).
>>
>> http://www60.zippyshare.com/v/13352963/file.html
> Please provide a direct link which doesn't require to enable scripting
> in a web browser to download.
>
>
> Thanks,
> Corinna
>
Please download this picture (jpg) and rename the file suffix to 7z or
open with corresponding archiver directly.

http://abload.de/img/patchfus75.jpg (99KB)


--
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: Bug with Cygwin's 'quilt' is actually in 'patch'

[Corinna Vinschen]

On Jul  3 15:04, Peter B. wrote:
> 
> On 03.07.2013 14:33 Corinna Vinschen wrote:
> > On Jul  3 05:03, Peter B. wrote:
> >> On Jun 20 09:46, Corinna Vinschen wrote:
> >>> I'll build a new patch 2.7.1 for 32 bit today. I hope that fixes it
> >>> for 32 bit as well.
> >> The new 2.7.1 (32bit) brakes CRLF handling for me completely. The old
> >> 2.6.1 works properly. 2.7.1 with text files converted from CRLF to LF
> >> works properly as well.
> >>
> >> Can you take a look at the download link below? It includes an example
> >> file (with CRLF), the corresponding patch (LF) and the two binaries
> >> (patch v2.6.1 and v2.7.1).
> >>
> >> http://www60.zippyshare.com/v/13352963/file.html
> > Please provide a direct link which doesn't require to enable scripting
> > in a web browser to download.
> >
> >
> > Thanks,
> > Corinna
> >
> Please download this picture (jpg) and rename the file suffix to 7z or
> open with corresponding archiver directly.
> 
> http://abload.de/img/patchfus75.jpg (99KB)

Thanks.  The behaviour here is a deliberate upstream choice to apply the
patch only if the line endings are the same as in the file to patch.

If you really must apply non-CRLF patches to CRLF files, consider using
a temporary text mount:
http://cygwin.com/cygwin-ug-net/using-utils.html#mount


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 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: Bug with Cygwin's 'quilt' is actually in 'patch'

[Peter B.]

On 03.07.2013 15:30 Corinna Vinschen wrote:
> On Jul  3 15:04, Peter B. wrote:
>> On 03.07.2013 14:33 Corinna Vinschen wrote:
>>> On Jul  3 05:03, Peter B. wrote:
>>>> On Jun 20 09:46, Corinna Vinschen wrote:
>>>>> I'll build a new patch 2.7.1 for 32 bit today. I hope that fixes it
>>>>> for 32 bit as well.
>>>> The new 2.7.1 (32bit) brakes CRLF handling for me completely. The old
>>>> 2.6.1 works properly. 2.7.1 with text files converted from CRLF to LF
>>>> works properly as well.
>>>>
>>>> Can you take a look at the download link below? It includes an example
>>>> file (with CRLF), the corresponding patch (LF) and the two binaries
>>>> (patch v2.6.1 and v2.7.1).
>>>>
>>>> http://www60.zippyshare.com/v/13352963/file.html
>>> Please provide a direct link which doesn't require to enable scripting
>>> in a web browser to download.
>>>
>>>
>>> Thanks,
>>> Corinna
>>>
>> Please download this picture (jpg) and rename the file suffix to 7z or
>> open with corresponding archiver directly.
>>
>> http://abload.de/img/patchfus75.jpg (99KB)
> Thanks.  The behaviour here is a deliberate upstream choice to apply the
> patch only if the line endings are the same as in the file to patch.
>
> If you really must apply non-CRLF patches to CRLF files, consider using
> a temporary text mount:
> http://cygwin.com/cygwin-ug-net/using-utils.html#mount
>
>
> Corinna
>
Sounds like intended behavior. I will try to tackle the problem
otherwise then.

The culprit here is the JavaVM that produces different
EOLs on different platforms (running a binary decompiler -> txt).

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