Maintainer for git?

Maintainer for git?

[David Conrad]

Hi,

Is there currently a maintainer for the git package?

The cygwin version is at 1.7.9, which is from January 2012,
and the latest version is 1.8.5.2.

If the reason it hasn't been updated is that there is no maintainer,
I *may* be able to help create a newer package.

All the best,
David Conrad

--
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: Maintainer for git?

[Steven Penny]

On Fri, Jan 10, 2014 at 9:43 PM, David Conrad wrote
> Is there currently a maintainer for the git package?

I would say no

http://cygwin.com/ml/cygwin/2013-06/msg00489.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: Maintainer for git?

[Adam Dinwoodie]

On 11 January 2014 03:55, Steven Penny wrote:
> On Fri, Jan 10, 2014 at 9:43 PM, David Conrad wrote
>> Is there currently a maintainer for the git package?
>
> I would say no
>
> http://cygwin.com/ml/cygwin/2013-06/msg00489.html

Officially, the Git maintainer is Eric Blake. While he's been active on
the list relatively recently, I've not seen him respond to anything
about Git.

Prompted by these emails, I've resurrected my old plan to offer to take
over Git maintenance, and sent in an "Intent To Adopt" email earlier
today. I've not clear on what the process is for this, or how long it
takes, but if the powers that be are happy for me to take over, I'll
hopefully be able to get a more up-to-date version online PDQ.

In the meantime, you could use the [Cygwin Ports] version (currently
v1.8.3.4), or download [my build] of v1.8.5.2  (just extract the
relevant packages into the Cygwin root directory). Obviously neither of
these options are supported on this list.

[Cygwin Ports]: http://cygwinports.org/
[my build]: http://tastycake.net/~adam/cygwin/

--
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: Maintainer for git?

[Steven Penny]

On Sat, Jan 11, 2014 at 5:52 PM, Adam Dinwoodie wrote
> [my build]: http://tastycake.net/~adam/cygwin/

Does your build have issues with cygcrypto re

http://stackoverflow.com/a/21049635

--
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: Maintainer for git?

[Adam Dinwoodie]

On 12 January 2014 00:05, Steven Penny wrote:
> On Sat, Jan 11, 2014 at 5:52 PM, Adam Dinwoodie wrote
>> [my build]: http://tastycake.net/~adam/cygwin/
>
> Does your build have issues with cygcrypto re
>
> http://stackoverflow.com/a/21049635

I've cloned git://source.ffmpeg.org/ffmpeg four times each with my x86
and x86_64 builds, twice to full depth and twice with `--depth 916`.  I
haven't hit any errors.  Whatever the problem there was, it doesn't seem
to affect these builds.

--
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: Maintainer for git?

[Steven Penny]

On Sat, Jan 11, 2014 at 5:52 PM, Adam Dinwoodie wrote
> [my build]: http://tastycake.net/~adam/cygwin/

Sorry, no dice

# git --version
git version 1.8.5.2

# git push
fatal: Unable to find remote helper for 'https'

--
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: Maintainer for git?

[Adam Dinwoodie]

On 12 January 2014 02:58, Steven Penny wrote:
> On Sat, Jan 11, 2014 at 5:52 PM, Adam Dinwoodie wrote
>> [my build]: http://tastycake.net/~adam/cygwin/
>
> Sorry, no dice
>
> # git --version
> git version 1.8.5.2
>
> # git push
> fatal: Unable to find remote helper for 'https'

Huh. Damn. I'm getting the same attempting to clone over https. I'll go
digging now.

Thank you for testing and reporting!

--
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: Maintainer for git?

[Adam Dinwoodie]

On Sun, Jan 12, 2014 at 01:33:17PM +0000, Adam Dinwoodie wrote:
> On 12 January 2014 02:58, Steven Penny wrote:
> > On Sat, Jan 11, 2014 at 5:52 PM, Adam Dinwoodie wrote
> >> [my build]: http://tastycake.net/~adam/cygwin/
> >
> > Sorry, no dice
> >
> > # git --version
> > git version 1.8.5.2
> >
> > # git push
> > fatal: Unable to find remote helper for 'https'
> 
> Huh. Damn. I'm getting the same attempting to clone over https. I'll go
> digging now.
> 
> Thank you for testing and reporting!

This should now be fixed, with a new version uploaded to the same place.
If you'd care to try again I'd be very grateful!

(In case anyone's interested, it appears that if Git is compiled without
openssl-devel or equivalent installed, it will silently fail to compile
support for https connections.)

--
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: Maintainer for git?

[David Conrad]

On Wed, Jan 15, 2014 at 5:55 AM, Adam Dinwoodie <adam@dinwoodie.org> wrote:
> On Sun, Jan 12, 2014 at 01:33:17PM +0000, Adam Dinwoodie wrote:
>> On 12 January 2014 02:58, Steven Penny wrote:
>> > . . .
>> > # git push
>> > fatal: Unable to find remote helper for 'https'
>>
>> . . .
>> Thank you for testing and reporting!
>
> This should now be fixed, with a new version uploaded to the same place.
> If you'd care to try again I'd be very grateful!
>
> (In case anyone's interested, it appears that if Git is compiled without
> openssl-devel or equivalent installed, it will silently fail to compile
> support for https connections.)

It sounds like we're quite close to having a new version available? I
know there's
one on Cygwin Ports or I could build it myself, of course, but I'm hoping to see
a new official build soon. Not to prod, but . . . *prod*. :)

Thanks for all you do (all of you, Cygwin folks),
David

--
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: Maintainer for git?

[Steven Penny]

On Wed, Jan 15, 2014 at 4:55 AM, Adam Dinwoodie wrote
>> > # git push
>> > fatal: Unable to find remote helper for 'https'
> This should now be fixed, with a new version uploaded to the same place.
> If you'd care to try again I'd be very grateful!

Yes sir, it is fixed! Sorry for the late reply.

--
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: Maintainer for git?

[Balaji Venkataraman]

On Mon, Jan 20, 2014 at 3:26 PM, Steven Penny wrote:
> On Wed, Jan 15, 2014 at 4:55 AM, Adam Dinwoodie wrote
>>> > # git push
>>> > fatal: Unable to find remote helper for 'https'
>> This should now be fixed, with a new version uploaded to the same place.
>> If you'd care to try again I'd be very grateful!
>
> Yes sir, it is fixed! Sorry for the late reply.

But the cygcrypto dll version bug referred here[1] is still present as
this version (1.8.5.2-1) is linked w/ cygcrypto-1.0.0.dll

[1] http://cygwin.com/ml/cygwin/2014-01/msg00134.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: Maintainer for git?

[Steven Penny]

On Mon, Jan 20, 2014 at 7:26 PM, Balaji Venkataraman wrote
> But the cygcrypto dll version bug referred here[1] is still present as
> this version (1.8.5.2-1) is linked w/ cygcrypto-1.0.0.dll

Are you certain the bug is present in Adam’s version? He has tested it here

    http://cygwin.com/ml/cygwin/2014-01/msg00085.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: Maintainer for git?

[David Conrad]

On Mon, Jan 20, 2014 at 8:31 PM, Steven Penny wrote:
> On Mon, Jan 20, 2014 at 7:26 PM, Balaji Venkataraman wrote
>> But the cygcrypto dll version bug referred here[1] is still present as
>> this version (1.8.5.2-1) is linked w/ cygcrypto-1.0.0.dll
>
> Are you certain the bug is present in Adam’s version? He has tested it here
>
>     http://cygwin.com/ml/cygwin/2014-01/msg00085.html

I just wanted to add, I have been using Adam's version for the last
couple of days without any issues. (The x86 variant.) The only problem
I had was when I went to install it using Steven's instructions:

> Also for anyone interested in testing this is the install, pretty simple
>
>     wget tastycake.net/~adam/cygwin/x86/git/git-1.8.5.2-1.tar.xz
>     tar -x -C / -f git-1.8.5.2-1.tar.xz

When I went to untar it, I got errors trying to hardlink git.exe and
git-upload-archive.exe to git-receive-pack.exe in the /usr/bin
directory. (I had previously uninstalled git using setup-x86.exe, so
as to install it "clean".) I ended up running Windows cmd.exe and
using mklink /H to make the links.

David

--
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: Maintainer for git?

[Adam Dinwoodie]

On 22 January 2014 18:25, David Conrad wrote:
> On Mon, Jan 20, 2014 at 8:31 PM, Steven Penny wrote:
> I just wanted to add, I have been using Adam's version for the last
> couple of days without any issues. (The x86 variant.) The only problem
> I had was when I went to install it using Steven's instructions:
>
>> Also for anyone interested in testing this is the install, pretty simple
>>
>>     wget tastycake.net/~adam/cygwin/x86/git/git-1.8.5.2-1.tar.xz
>>     tar -x -C / -f git-1.8.5.2-1.tar.xz
>
> When I went to untar it, I got errors trying to hardlink git.exe and
> git-upload-archive.exe to git-receive-pack.exe in the /usr/bin
> directory. (I had previously uninstalled git using setup-x86.exe, so
> as to install it "clean".) I ended up running Windows cmd.exe and
> using mklink /H to make the links.

I've not had any issues I've noticed with hard links using my build, and
I've been fetching using the git:// protocol; I'm pretty sure that uses
git-receive-pack.

I want to try performing a minimal installation next, to check I've got
the dependency list correct, so that may highlight what the problem here
is.

The only difference between my process and the one Steven suggested is I
added a -a switch to the tar call, but that seems to be redundant 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: Maintainer for git?

[Balaji Venkataraman]

On Mon, Jan 20, 2014 at 5:31 PM, Steven Penny wrote:

> Are you certain the bug is present in Adam’s version? He has tested it here
>
>     http://cygwin.com/ml/cygwin/2014-01/msg00085.html

I just tried cloning something big again (the linux kernel). See
below. The first one was with git-1.7.9 the default Cygwin version.

x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
linux-git-1.7.9
Cloning into 'linux-git-1.7.9'...
remote: Counting objects: 3366700, done.
remote: Compressing objects: 100% (507909/507909), done.
remote: Total 3366700 (delta 2833548), reused 3362709 (delta 2829741)
Receiving objects: 100% (3366700/3366700), 705.04 MiB | 3.25 MiB/s, done.
Resolving deltas: 100% (2833548/2833548), done.

x86$ cd /

x86$ tar xf ~/git-1.8.5.2-1.tar.xz

x86$ cd

x86$ git --version
git version 1.8.5.2

x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
linux-git-1.8.5.2
Cloning into 'linux-git-1.8.5.2'...
remote: Counting objects: 3366700, done.
remote: Compressing objects: 100% (508113/508113), done.
remote: Total 3366700 (delta 2833563), reused 3362434 (delta 2829537)
Receiving objects: 100% (3366700/3366700), 703.64 MiB | 2.16 MiB/s, done.
fatal: pack is corrupted (SHA1 mismatch)
fatal: index-pack failed

--
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: Maintainer for git?

[Chris O'Bryan]

> On Mon, Jan 20, 2014 at 5:31 PM, Steven Penny wrote:
>
>> Are you certain the bug is present in Adam’s version? He has tested it here
>>
>>     http://cygwin.com/ml/cygwin/2014-01/msg00085.html
>
> I just tried cloning something big again (the linux kernel). See
> below. The first one was with git-1.7.9 the default Cygwin version.
>
> x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
> linux-git-1.7.9
> Cloning into 'linux-git-1.7.9'...
> remote: Counting objects: 3366700, done.
> remote: Compressing objects: 100% (507909/507909), done.
> remote: Total 3366700 (delta 2833548), reused 3362709 (delta 2829741)
> Receiving objects: 100% (3366700/3366700), 705.04 MiB | 3.25 MiB/s, done.
> Resolving deltas: 100% (2833548/2833548), done.
>
> x86$ cd /
>
> x86$ tar xf ~/git-1.8.5.2-1.tar.xz
>
> x86$ cd
>
> x86$ git --version
> git version 1.8.5.2
>
> x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
> linux-git-1.8.5.2
> Cloning into 'linux-git-1.8.5.2'...
> remote: Counting objects: 3366700, done.
> remote: Compressing objects: 100% (508113/508113), done.
> remote: Total 3366700 (delta 2833563), reused 3362434 (delta 2829537)
> Receiving objects: 100% (3366700/3366700), 703.64 MiB | 2.16 MiB/s, done.
> fatal: pack is corrupted (SHA1 mismatch)
> fatal: index-pack failed
>
Could you try rerunning that same clone again with 1.8.5.2, but first
temporarily replace cygcrypto-1.0.0.dll with a copy of
cygcrypto-0.9.8.dll ? If it works with the only variable being the
version of cygcrypto then I think that's pretty strong evidence that
there is some regression in cygcrypto. It's odd that it is
intermittent, though. Does anyone know anything about cygcrypto? Are
there functions that maybe aren't reentrant? The only reasons I can
think of for it sometimes working are either threading issues, using
uninitialized memory, or memory corruption (buffer overflow somewhere
perhaps).

Chris

--
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: Maintainer for git?

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 2698 bytes --]

On Jan 22 16:42, Chris O'Bryan wrote:
> > On Mon, Jan 20, 2014 at 5:31 PM, Steven Penny wrote:
> >
> >> Are you certain the bug is present in Adam’s version? He has tested it here
> >>
> >>     http://cygwin.com/ml/cygwin/2014-01/msg00085.html
> >
> > I just tried cloning something big again (the linux kernel). See
> > below. The first one was with git-1.7.9 the default Cygwin version.
> >
> > x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
> > linux-git-1.7.9
> > Cloning into 'linux-git-1.7.9'...
> > remote: Counting objects: 3366700, done.
> > remote: Compressing objects: 100% (507909/507909), done.
> > remote: Total 3366700 (delta 2833548), reused 3362709 (delta 2829741)
> > Receiving objects: 100% (3366700/3366700), 705.04 MiB | 3.25 MiB/s, done.
> > Resolving deltas: 100% (2833548/2833548), done.
> >
> > x86$ cd /
> >
> > x86$ tar xf ~/git-1.8.5.2-1.tar.xz
> >
> > x86$ cd
> >
> > x86$ git --version
> > git version 1.8.5.2
> >
> > x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
> > linux-git-1.8.5.2
> > Cloning into 'linux-git-1.8.5.2'...
> > remote: Counting objects: 3366700, done.
> > remote: Compressing objects: 100% (508113/508113), done.
> > remote: Total 3366700 (delta 2833563), reused 3362434 (delta 2829537)
> > Receiving objects: 100% (3366700/3366700), 703.64 MiB | 2.16 MiB/s, done.
> > fatal: pack is corrupted (SHA1 mismatch)
> > fatal: index-pack failed
> >
> Could you try rerunning that same clone again with 1.8.5.2, but first
> temporarily replace cygcrypto-1.0.0.dll with a copy of
> cygcrypto-0.9.8.dll ? If it works with the only variable being the
> version of cygcrypto then I think that's pretty strong evidence that
> there is some regression in cygcrypto. It's odd that it is
> intermittent, though. Does anyone know anything about cygcrypto? Are
> there functions that maybe aren't reentrant? The only reasons I can
> think of for it sometimes working are either threading issues, using
> uninitialized memory, or memory corruption (buffer overflow somewhere
> perhaps).

Even if it's an OpenSSL issue, I'm not sure what to do about it.  I'm
building OpenSSL for Cygwin, but I have no knowledge of the crypto
stuff...

I don't see anything unusual in the build process, except, maybe, that
for SHA1 the x86 assembler code is used.  I'm wondering what happens
when taking out -DSHA1_ASM from the build process.  Or what happens when
using -O2 instead of -O3.


Corinna

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

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

Re: Maintainer for git?

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 2172 bytes --]

On Jan 23 11:51, Corinna Vinschen wrote:
> On Jan 22 16:42, Chris O'Bryan wrote:
> > > On Mon, Jan 20, 2014 at 5:31 PM, Steven Penny wrote:
> > >
> > >> Are you certain the bug is present in Adam’s version? He has tested it here
> > >>
> > >>     http://cygwin.com/ml/cygwin/2014-01/msg00085.html
> > >
> > > I just tried cloning something big again (the linux kernel). See
> > > below. The first one was with git-1.7.9 the default Cygwin version.
> > >[...]
> > > x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
> > > linux-git-1.8.5.2
> > > Cloning into 'linux-git-1.8.5.2'...
> > > remote: Counting objects: 3366700, done.
> > > remote: Compressing objects: 100% (508113/508113), done.
> > > remote: Total 3366700 (delta 2833563), reused 3362434 (delta 2829537)
> > > Receiving objects: 100% (3366700/3366700), 703.64 MiB | 2.16 MiB/s, done.
> > > fatal: pack is corrupted (SHA1 mismatch)
> > > fatal: index-pack failed
> > >[...]
> 
> Even if it's an OpenSSL issue, I'm not sure what to do about it.  I'm
> building OpenSSL for Cygwin, but I have no knowledge of the crypto
> stuff...
> 
> I don't see anything unusual in the build process, except, maybe, that
> for SHA1 the x86 assembler code is used.  I'm wondering what happens
> when taking out -DSHA1_ASM from the build process.  Or what happens when
> using -O2 instead of -O3.

I have a hunch that this is not a OpenSSL but a Cygwin problem.

The default case in OpenSSL is to use SSE2 instructions in the x86
assembler code.  However, SSE2 instructions utilize the x86 XMM
registers, which are not saved and restored in setjmp/longjmp, nor are
they stored and restored in signal handling under Cygwin.

In the long run Cygwin should save and restore the XMM registers on
x86 as well, I guess.

For the time being, I've build a new OpenSSL version 1.0.1f-2 with the
"no-sse2" flag.  With this version I could clone the linx repo without
error.  Please give it a try.


Corinna

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

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

Re: Maintainer for git?

[David Rothenberger]

On 1/25/2014 3:59 AM, Corinna Vinschen wrote:
> On Jan 23 11:51, Corinna Vinschen wrote:
>> On Jan 22 16:42, Chris O'Bryan wrote:
>>>> On Mon, Jan 20, 2014 at 5:31 PM, Steven Penny wrote:
>>>> 
>>>>> Are you certain the bug is present in AdamÂ’s version? He
>>>>> has tested it here
>>>>> 
>>>>> http://cygwin.com/ml/cygwin/2014-01/msg00085.html
>>>> 
>>>> I just tried cloning something big again (the linux kernel).
>>>> See below. The first one was with git-1.7.9 the default
>>>> Cygwin version. [...] x86$ git clone
>>>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
>>>>
>>>> 
linux-git-1.8.5.2
>>>> Cloning into 'linux-git-1.8.5.2'... remote: Counting objects:
>>>> 3366700, done. remote: Compressing objects: 100%
>>>> (508113/508113), done. remote: Total 3366700 (delta 2833563),
>>>> reused 3362434 (delta 2829537) Receiving objects: 100%
>>>> (3366700/3366700), 703.64 MiB | 2.16 MiB/s, done. fatal: pack
>>>> is corrupted (SHA1 mismatch) fatal: index-pack failed [...]
>> 
>> Even if it's an OpenSSL issue, I'm not sure what to do about it.
>> I'm building OpenSSL for Cygwin, but I have no knowledge of the
>> crypto stuff...
>> 
>> I don't see anything unusual in the build process, except, maybe,
>> that for SHA1 the x86 assembler code is used.  I'm wondering what
>> happens when taking out -DSHA1_ASM from the build process.  Or
>> what happens when using -O2 instead of -O3.
> 
> I have a hunch that this is not a OpenSSL but a Cygwin problem.
> 
> The default case in OpenSSL is to use SSE2 instructions in the x86 
> assembler code.  However, SSE2 instructions utilize the x86 XMM 
> registers, which are not saved and restored in setjmp/longjmp, nor
> are they stored and restored in signal handling under Cygwin.
> 
> In the long run Cygwin should save and restore the XMM registers
> on x86 as well, I guess.
> 
> For the time being, I've build a new OpenSSL version 1.0.1f-2 with
> the "no-sse2" flag.  With this version I could clone the linx repo
> without error.  Please give it a try.

The new OpenSSL version fixes things for me. This explanation makes
sense given what I've seen as well.

I had built my own updated git binaries a while ago. I noticed that
for x86, the binaries would fail cloning a large repository, and would
also give spurious errors from "git fsck" on an existing repository. I
had no such problems with my x86_64 build.

With the new OpenSSL on x86, both the clone and the fsck problems have
gone away.

Thanks! Now my colleagues can use Git 1.8 with x86!


-- 
David Rothenberger  ----  daveroth@acm.org

Revolution, n.:
        In politics, an abrupt change in the form of misgovernment.
                -- Ambrose Bierce

--
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: Maintainer for git?

[Balaji Venkataraman]

On Sat, Jan 25, 2014 at 3:59 AM, Corinna Vinschen wrote:
> I have a hunch that this is not a OpenSSL but a Cygwin problem.
>
> The default case in OpenSSL is to use SSE2 instructions in the x86
> assembler code.  However, SSE2 instructions utilize the x86 XMM
> registers, which are not saved and restored in setjmp/longjmp, nor are
> they stored and restored in signal handling under Cygwin.
>
> In the long run Cygwin should save and restore the XMM registers on
> x86 as well, I guess.
>
> For the time being, I've build a new OpenSSL version 1.0.1f-2 with the
> "no-sse2" flag.  With this version I could clone the linx repo without
> error.  Please give it a try.

Thanks Corinna for the openssl updates. I definitely see an
improvement with them. I was able to clone the ffmpeg repo w/o errors
- which was not happening earlier. However, the linux repo clone
failed w/ the same error as before. Not sure if there is something
else that could be going on.

x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
linux
Cloning into 'linux'...
remote: Counting objects: 3407728, done.
remote: Compressing objects: 100% (517899/517899), done.
remote: Total 3407728 (delta 2869094), reused 3399100 (delta 2860672)
Receiving objects: 100% (3407728/3407728), 713.94 MiB | 619.00 KiB/s, done.
fatal: pack is corrupted (SHA1 mismatch)
fatal: index-pack failed

--
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: Maintainer for git?

[Adam Dinwoodie]

On Sat, Jan 25, 2014 at 04:55:57PM -0800, Balaji Venkataraman wrote:
> On Sat, Jan 25, 2014 at 3:59 AM, Corinna Vinschen wrote:
> > I have a hunch that this is not a OpenSSL but a Cygwin problem.
> >
> > The default case in OpenSSL is to use SSE2 instructions in the x86
> > assembler code.  However, SSE2 instructions utilize the x86 XMM
> > registers, which are not saved and restored in setjmp/longjmp, nor are
> > they stored and restored in signal handling under Cygwin.
> >
> > In the long run Cygwin should save and restore the XMM registers on
> > x86 as well, I guess.
> >
> > For the time being, I've build a new OpenSSL version 1.0.1f-2 with the
> > "no-sse2" flag.  With this version I could clone the linx repo without
> > error.  Please give it a try.
> 
> Thanks Corinna for the openssl updates. I definitely see an
> improvement with them. I was able to clone the ffmpeg repo w/o errors
> - which was not happening earlier.

This is good to hear, particularly given I was unable to reproduce that
problem.

> However, the linux repo clone
> failed w/ the same error as before. Not sure if there is something
> else that could be going on.
> 
> x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
> linux
> Cloning into 'linux'...
> remote: Counting objects: 3407728, done.
> remote: Compressing objects: 100% (517899/517899), done.
> remote: Total 3407728 (delta 2869094), reused 3399100 (delta 2860672)
> Receiving objects: 100% (3407728/3407728), 713.94 MiB | 619.00 KiB/s, done.
> fatal: pack is corrupted (SHA1 mismatch)
> fatal: index-pack failed

I was previously managing to reproduce this, but since updating to
Corinna's latest OpenSSL builds, I've not had any issues.  Is it
possibly an out-of-date library hanging around in memory?  Have you
tried restarting all your Cygwin processes and/or rebooting your system?
I'm surprised Corinna's changes would fix it for me but not for you.

--
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: Maintainer for git?

[Balaji Venkataraman]

On Sun, Jan 26, 2014 at 11:19 AM, Adam Dinwoodie wrote:

> I was previously managing to reproduce this, but since updating to
> Corinna's latest OpenSSL builds, I've not had any issues.  Is it
> possibly an out-of-date library hanging around in memory?  Have you
> tried restarting all your Cygwin processes and/or rebooting your system?
> I'm surprised Corinna's changes would fix it for me but not for you.

Based on the above, I went back and checked if the openssl update was
all good. Not quite. Apparently, since I had sshd running when I
updated the openssl package, looks like the cygcrypto-1.0.0.dll was
not properly replaced. I ended up w/ a .new file (created by
setup.exe(?)). I killed sshd renamed the .dll and reran the clone. It
seems to work (see below) except for one error but I do have a clone
now. It might be a good idea to perhaps print some warning messages
when updating packages whose dlls are in use - unless that is already
happening and I have missed them. But overall, the fix seems to do the
trick. Thanks again to all.

x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
linux
Cloning into 'linux'...
remote: Counting objects: 3408476, done.
remote: Compressing objects: 100% (517865/517865), done.
Receiving objects: 100% (3408476/3408476), 712.95 MiB | 460.00 KiB/s, done.
remote: Total 3408476 (delta 2869731), reused 3400091 (delta 2861452)
Resolving deltas: 100% (2869731/2869731), done.
Checking connectivity... done.
error: git-remote-https died of signal 13
Checking out files: 100% (45625/45625), done.

--
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: Maintainer for git?

[Balaji Venkataraman]

On Mon, Jan 27, 2014 at 2:30 PM, Balaji Venkataraman wrote:
> On Sun, Jan 26, 2014 at 11:19 AM, Adam Dinwoodie wrote:
>
>> I was previously managing to reproduce this, but since updating to
>> Corinna's latest OpenSSL builds, I've not had any issues.  Is it
>> possibly an out-of-date library hanging around in memory?  Have you
>> tried restarting all your Cygwin processes and/or rebooting your system?
>> I'm surprised Corinna's changes would fix it for me but not for you.
>
> Based on the above, I went back and checked if the openssl update was
> all good. Not quite. Apparently, since I had sshd running when I
> updated the openssl package, looks like the cygcrypto-1.0.0.dll was
> not properly replaced. I ended up w/ a .new file (created by
> setup.exe(?)). I killed sshd renamed the .dll and reran the clone. It
> seems to work (see below) except for one error but I do have a clone
> now. It might be a good idea to perhaps print some warning messages
> when updating packages whose dlls are in use - unless that is already
> happening and I have missed them. But overall, the fix seems to do the
> trick. Thanks again to all.
>
> x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
> linux
> Cloning into 'linux'...
> remote: Counting objects: 3408476, done.
> remote: Compressing objects: 100% (517865/517865), done.
> Receiving objects: 100% (3408476/3408476), 712.95 MiB | 460.00 KiB/s, done.
> remote: Total 3408476 (delta 2869731), reused 3400091 (delta 2861452)
> Resolving deltas: 100% (2869731/2869731), done.
> Checking connectivity... done.
> error: git-remote-https died of signal 13
> Checking out files: 100% (45625/45625), done.

I did a very quick search on the git-remote-https error and I found a
recent thread[1]. Hoping someone can track it down on Cygwin too.
FWIW, I don't have curl installed - just libcurl.

cygcheck -f /usr/bin/*curl*
libcurl4-7.34.0-1

[1] http://git.661346.n2.nabble.com/error-git-remote-https-died-of-signal-13-td7599771.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: Maintainer for git?

[Andrey Repin]

Greetings, Balaji Venkataraman!

> I did a very quick search on the git-remote-https error and I found a
> recent thread[1]. Hoping someone can track it down on Cygwin too.
> FWIW, I don't have curl installed - just libcurl.

libcurl IS THE cURL. The library. The "curl" you are referring to is just a
console front-end to it.

> cygcheck -f /usr/bin/*curl*
> libcurl4-7.34.0-1


--
WBR,
Andrey Repin (anrdaemon@yandex.ru) 28.01.2014, <03:26>

Sorry for my terrible english...


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

Re: Maintainer for git?

[Larry Hall (Cygwin)]

On 1/27/2014 5:30 PM, Balaji Venkataraman wrote:
> On Sun, Jan 26, 2014 at 11:19 AM, Adam Dinwoodie wrote:
>
>> I was previously managing to reproduce this, but since updating to
>> Corinna's latest OpenSSL builds, I've not had any issues.  Is it
>> possibly an out-of-date library hanging around in memory?  Have you
>> tried restarting all your Cygwin processes and/or rebooting your system?
>> I'm surprised Corinna's changes would fix it for me but not for you.
>
> Based on the above, I went back and checked if the openssl update was
> all good. Not quite. Apparently, since I had sshd running when I
> updated the openssl package, looks like the cygcrypto-1.0.0.dll was
> not properly replaced. I ended up w/ a .new file (created by
> setup.exe(?)).

Yes, this is what setup*.exe does if the DLL it's trying to replace is
in use.  setup*.exe will warn you about such cases and ask if you want
to continue (and have to reboot to complete the installation) or if you
prefer to kill the app using the DLL.  You apparently missed that and
ended up with the former option.

> I killed sshd renamed the .dll and reran the clone. It
> seems to work (see below) except for one error but I do have a clone
> now.

FYI, this is not the recommended way to handle this, since it leaves
a scheduled DLL move on reboot in an unpredictable state.  So if you
see a DLL with this extension in the future, just reboot.

> It might be a good idea to perhaps print some warning messages
> when updating packages whose dlls are in use - unless that is already
> happening and I have missed them.

Right, see above.


-- 
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: Maintainer for git?

[Steven Penny]

On Sat, Jan 25, 2014 at 5:59 AM, Corinna Vinschen wrote
> For the time being, I've build a new OpenSSL version 1.0.1f-2 with the
> "no-sse2" flag.  With this version I could clone the linx repo without
> error.  Please give it a try.

I can confirm this fixes my problems as well

    $ git clone git://github.com/ruby/ruby
    Cloning into 'ruby'...
    remote: Reusing existing pack: 277333, done.
    remote: Counting objects: 6, done.
    remote: Compressing objects: 100% (5/5), done.
    remote: Total 277339 (delta 1), reused 6 (delta 1)
    Receiving objects: 100% (277339/277339), 140.00 MiB | 1.49 MiB/s, done.
    Resolving deltas: 100% (214277/214277), done.
    Checking connectivity... done.
    Checking out files: 100% (4217/4217), done.

--
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: Maintainer for git?

[Chris O'Bryan]

On Sun, Jan 26, 2014 at 4:45 PM, Steven Penny wrote:
> On Sat, Jan 25, 2014 at 5:59 AM, Corinna Vinschen wrote
>> For the time being, I've build a new OpenSSL version 1.0.1f-2 with the
>> "no-sse2" flag.  With this version I could clone the linx repo without
>> error.  Please give it a try.
>
> I can confirm this fixes my problems as well
>

This fixed my issue, too. Good hunch on the SSE stuff!

Thanks,
Chris

--
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: Maintainer for git?

[David Conrad]

On Mon, Jan 27, 2014 at 10:12 AM, Chris O'Bryan wrote:
> On Sun, Jan 26, 2014 at 4:45 PM, Steven Penny wrote:
>> On Sat, Jan 25, 2014 at 5:59 AM, Corinna Vinschen wrote
>>> For the time being, I've build a new OpenSSL version 1.0.1f-2 with the
>>> "no-sse2" flag.  With this version I could clone the linx repo without
>>> error.  Please give it a try.
>>
>> I can confirm this fixes my problems as well
>>
>
> This fixed my issue, too. Good hunch on the SSE stuff!
>

With the new OpenSSL from Corinna I was able to clone both the linux
and ffmpeg repos with Adam's latest git-1.8.5.2.

Many thanks to Corinna, Adam, and all those who have tested this. I
hope to see a new git on the mirrors soon!

Cheers,
David

--
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: Maintainer for git?

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 1864 bytes --]

Hey guys,

On Jan 27 16:14, David Conrad wrote:
> On Mon, Jan 27, 2014 at 10:12 AM, Chris O'Bryan wrote:
> > On Sun, Jan 26, 2014 at 4:45 PM, Steven Penny wrote:
> >> On Sat, Jan 25, 2014 at 5:59 AM, Corinna Vinschen wrote
> >>> For the time being, I've build a new OpenSSL version 1.0.1f-2 with the
> >>> "no-sse2" flag.  With this version I could clone the linx repo without
> >>> error.  Please give it a try.
> >>
> >> I can confirm this fixes my problems as well
> >>
> >
> > This fixed my issue, too. Good hunch on the SSE stuff!
> >
> 
> With the new OpenSSL from Corinna I was able to clone both the linux
> and ffmpeg repos with Adam's latest git-1.8.5.2.
> 
> Many thanks to Corinna, Adam, and all those who have tested this. I
> hope to see a new git on the mirrors soon!

many thanks to all of you for your feedback.

Chris and I discussed this problem further and we applied a patch to the
Cygwin DLL which saves and restores the FPU state and XMM registers on
32 bit as well when a thread gets interrupted for signal handling.

For testing I reverted my local OpenSSL to the previous 1.0.1f-1 and
tried to fetch the Linux git repo a couple of times without error, so
I'm carefully optimistic that we can revert to an SSE2-enabled OpenSSL
as soon as we release the next Cygwin version 1.7.28, which is due very
soon now.

I'd be grateful if you could do some more testing as well.  Please
revert your OpenSSL package to 1.0.1f-1 via setup, and then fetch the
latest 2014-01-28 Cygwin snapshot DLL from http://cygwin.com/snapshots/,

With this combination, you should be able to git clone the same as
with the SSE2-less OpenSSL.


Thanks in advance,
Corinna

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

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

Re: Maintainer for git?

[Adam Dinwoodie]

On Tue, Jan 28, 2014 at 12:11:02PM +0100, Corinna Vinschen wrote:
> Chris and I discussed this problem further and we applied a patch to the
> Cygwin DLL which saves and restores the FPU state and XMM registers on
> 32 bit as well when a thread gets interrupted for signal handling.
> 
> For testing I reverted my local OpenSSL to the previous 1.0.1f-1 and
> tried to fetch the Linux git repo a couple of times without error, so
> I'm carefully optimistic that we can revert to an SSE2-enabled OpenSSL
> as soon as we release the next Cygwin version 1.7.28, which is due very
> soon now.
> 
> I'd be grateful if you could do some more testing as well.  Please
> revert your OpenSSL package to 1.0.1f-1 via setup, and then fetch the
> latest 2014-01-28 Cygwin snapshot DLL from http://cygwin.com/snapshots/,
> 
> With this combination, you should be able to git clone the same as
> with the SSE2-less OpenSSL.

I've just tested this -- the clone is failing as expected with the
current cygwin package from setup-x86.exe, and working just fine with
the 2014-01-28 snapshot.  I've not tested the snapshot on x86_64, but I
can try to find time to do that if it'd be useful.

I have to say, I'm really impressed with your debugging 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: Maintainer for git?

[David Conrad]

On Wed, Jan 29, 2014 at 10:35 AM, Adam Dinwoodie wrote:
> On Tue, Jan 28, 2014 at 12:11:02PM +0100, Corinna Vinschen wrote:
>> Chris and I discussed this problem further and we applied a patch to the
>> Cygwin DLL which saves and restores the FPU state and XMM registers on
>> 32 bit as well when a thread gets interrupted for signal handling.
>> . . .
>> I'd be grateful if you could do some more testing as well.  Please
>> revert your OpenSSL package to 1.0.1f-1 via setup, and then fetch the
>> latest 2014-01-28 Cygwin snapshot DLL from http://cygwin.com/snapshots/,
>> . . .
>
> I've just tested this -- the clone is failing ... with the current cygwin
> package ... and working just fine with the 2014-01-28 snapshot. ...

I tried to test this as well, but when I tried to move the existing
cygwin1.dll out of the way and drop the new one in, I got "Access is
denied.", even when running a cmd.exe Windows shell as Administrator,
without any cygwin bash shell running, and ditto for trying it after a
reboot and not having run anything cygwin since the boot.

I'm not sure what Vulcan nerve pinch I'm missing here to convince
Windows to let me at these files.

> I have to say, I'm really impressed with your debugging here!

Ditto!

All the best,
David

--
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: Maintainer for git?

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 1471 bytes --]

On Jan 29 12:13, David Conrad wrote:
> On Wed, Jan 29, 2014 at 10:35 AM, Adam Dinwoodie wrote:
> > On Tue, Jan 28, 2014 at 12:11:02PM +0100, Corinna Vinschen wrote:
> >> Chris and I discussed this problem further and we applied a patch to the
> >> Cygwin DLL which saves and restores the FPU state and XMM registers on
> >> 32 bit as well when a thread gets interrupted for signal handling.
> >> . . .
> >> I'd be grateful if you could do some more testing as well.  Please
> >> revert your OpenSSL package to 1.0.1f-1 via setup, and then fetch the
> >> latest 2014-01-28 Cygwin snapshot DLL from http://cygwin.com/snapshots/,
> >> . . .
> >
> > I've just tested this -- the clone is failing ... with the current cygwin
> > package ... and working just fine with the 2014-01-28 snapshot. ...
> 
> I tried to test this as well, but when I tried to move the existing
> cygwin1.dll out of the way and drop the new one in, I got "Access is
> denied.", even when running a cmd.exe Windows shell as Administrator,
> without any cygwin bash shell running, and ditto for trying it after a
> reboot and not having run anything cygwin since the boot.
> 
> I'm not sure what Vulcan nerve pinch I'm missing here to convince
> Windows to let me at these files.

Still a service running, perhaps?


Corinna

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

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

Re: Maintainer for git?

[David Conrad]

On Wed, Jan 29, 2014 at 1:10 PM, Corinna Vinschen wrote:
> On Jan 29 12:13, David Conrad wrote:
>> On Wed, Jan 29, 2014 at 10:35 AM, Adam Dinwoodie wrote:
>> > On Tue, Jan 28, 2014 at 12:11:02PM +0100, Corinna Vinschen wrote:
>> >> Chris and I discussed this problem further and we applied a patch to the
>> >> Cygwin DLL which saves and restores the FPU state and XMM registers on
>> >> 32 bit as well when a thread gets interrupted for signal handling.
>> >> . . .
>> >> I'd be grateful if you could do some more testing as well.  Please
>> >> revert your OpenSSL package to 1.0.1f-1 via setup, and then fetch the
>> >> latest 2014-01-28 Cygwin snapshot DLL from http://cygwin.com/snapshots/,
>> >> . . .
>> >
>> > I've just tested this -- the clone is failing ... with the current cygwin
>> > package ... and working just fine with the 2014-01-28 snapshot. ...

Well, one way to test with the new Cygwin snapshot DLL is just to wait
until it becomes the new release. :)

I just tested git cloning ffmpeg and linux, with Adam's git 1.8.5.2,
on 32-bit Cygwin 1.7.28, and using the prior version of the OpenSSL
package, 1.0.1f-1. It worked like a charm. I'm guessing the plan is to
roll out a "new" OpenSSL package that is compiled with SSE2 enabled
again, and to release the updated build of git?

David

--
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: Maintainer for git?

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 1667 bytes --]

On Feb  4 16:10, David Conrad wrote:
> On Wed, Jan 29, 2014 at 1:10 PM, Corinna Vinschen wrote:
> > On Jan 29 12:13, David Conrad wrote:
> >> On Wed, Jan 29, 2014 at 10:35 AM, Adam Dinwoodie wrote:
> >> > On Tue, Jan 28, 2014 at 12:11:02PM +0100, Corinna Vinschen wrote:
> >> >> Chris and I discussed this problem further and we applied a patch to the
> >> >> Cygwin DLL which saves and restores the FPU state and XMM registers on
> >> >> 32 bit as well when a thread gets interrupted for signal handling.
> >> >> . . .
> >> >> I'd be grateful if you could do some more testing as well.  Please
> >> >> revert your OpenSSL package to 1.0.1f-1 via setup, and then fetch the
> >> >> latest 2014-01-28 Cygwin snapshot DLL from http://cygwin.com/snapshots/,
> >> >> . . .
> >> >
> >> > I've just tested this -- the clone is failing ... with the current cygwin
> >> > package ... and working just fine with the 2014-01-28 snapshot. ...
> 
> Well, one way to test with the new Cygwin snapshot DLL is just to wait
> until it becomes the new release. :)
> 
> I just tested git cloning ffmpeg and linux, with Adam's git 1.8.5.2,
> on 32-bit Cygwin 1.7.28, and using the prior version of the OpenSSL
> package, 1.0.1f-1. It worked like a charm. 

Thanks for testing!

> I'm guessing the plan is to
> roll out a "new" OpenSSL package that is compiled with SSE2 enabled
> again, and to release the updated build of git?

Yes, that's the idea.  I'll do that in the next couple of days.


Corinna

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

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

Re: Maintainer for git?

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 1645 bytes --]

On Jan 29 15:35, Adam Dinwoodie wrote:
> On Tue, Jan 28, 2014 at 12:11:02PM +0100, Corinna Vinschen wrote:
> > Chris and I discussed this problem further and we applied a patch to the
> > Cygwin DLL which saves and restores the FPU state and XMM registers on
> > 32 bit as well when a thread gets interrupted for signal handling.
> > 
> > For testing I reverted my local OpenSSL to the previous 1.0.1f-1 and
> > tried to fetch the Linux git repo a couple of times without error, so
> > I'm carefully optimistic that we can revert to an SSE2-enabled OpenSSL
> > as soon as we release the next Cygwin version 1.7.28, which is due very
> > soon now.
> > 
> > I'd be grateful if you could do some more testing as well.  Please
> > revert your OpenSSL package to 1.0.1f-1 via setup, and then fetch the
> > latest 2014-01-28 Cygwin snapshot DLL from http://cygwin.com/snapshots/,
> > 
> > With this combination, you should be able to git clone the same as
> > with the SSE2-less OpenSSL.
> 
> I've just tested this -- the clone is failing as expected with the
> current cygwin package from setup-x86.exe, and working just fine with
> the 2014-01-28 snapshot.  I've not tested the snapshot on x86_64, but I
> can try to find time to do that if it'd be useful.

Thanks for testing.  This affects the 32 bit version in the first place,
so there's no rush to test it on 64 bit, afaics.

> I have to say, I'm really impressed with your debugging here!

Thanks :)


Corinna

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

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

Re: Maintainer for git?

[David Conrad]

On Wed, Jan 29, 2014 at 10:35 AM, Adam Dinwoodie wrote:
> On Tue, Jan 28, 2014 at 12:11:02PM +0100, Corinna Vinschen wrote:
>> as soon as we release the next Cygwin version 1.7.28, which is due very
>> soon now.
>>
> I've just tested this -- the clone is [...] working just fine with
> the 2014-01-28 snapshot.  I've not tested the snapshot on x86_64, but I
> can try to find time to do that if it'd be useful.
>

Just curious where we stand with this now that 1.7.28 is out. Does it
still need testing on x86_64? I currently have 32-bit cygwin installed
on a 64-bit machine, but I'd be happy to install 64 (they can be
installed side-by-side, correct?) if it would help with testing the
new git.

Regards,
David

--
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: Maintainer for git?

[Adam Dinwoodie]

On Mon, Feb 10, 2014 at 06:35:32PM -0500, David Conrad DavidConrad-at-acm.org |Cygwin| wrote:
> On Wed, Jan 29, 2014 at 10:35 AM, Adam Dinwoodie wrote:
> > On Tue, Jan 28, 2014 at 12:11:02PM +0100, Corinna Vinschen wrote:
> >> as soon as we release the next Cygwin version 1.7.28, which is due very
> >> soon now.
> >>
> > I've just tested this -- the clone is [...] working just fine with
> > the 2014-01-28 snapshot.  I've not tested the snapshot on x86_64, but I
> > can try to find time to do that if it'd be useful.
> 
> Just curious where we stand with this now that 1.7.28 is out. Does it
> still need testing on x86_64? I currently have 32-bit cygwin installed
> on a 64-bit machine, but I'd be happy to install 64 (they can be
> installed side-by-side, correct?) if it would help with testing the
> new git.

You can definitely install 32-bit and 64-bit alongside each other,
although if you do anything manually with your path, for example, you
should be careful the installs don't interfere with each other.

I've been using my Git install day-to-day on 32-bit Cygwin, mostly
because that's where I have most of my environment set up, but I've been
trying to test both when I've been explicitly testing my builds.

Adam

--
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: Maintainer for git?

[David Boyce]

On Tue, Feb 11, 2014 at 9:58 AM, Adam Dinwoodie <adam@dinwoodie.org> wrote:
>> Just curious where we stand with this now that 1.7.28 is out.

Assuming that further testing affirms that the SSE2 bug affecting Git
is fixed in recent snapshots, this brings us back the original
subject: is there a maintainer for Git who will package up the latest
release such that it becomes the new default? That topic seems to have
gotten lost along the way somehow.

David

--
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: Maintainer for git?

[Larry Hall (Cygwin)]

On 2/11/2014 2:03 PM, David Boyce wrote:
> On Tue, Feb 11, 2014 at 9:58 AM, Adam Dinwoodie wrote:
>>> Just curious where we stand with this now that 1.7.28 is out.
>
> Assuming that further testing affirms that the SSE2 bug affecting Git
> is fixed in recent snapshots, this brings us back the original
> subject: is there a maintainer for Git who will package up the latest
> release such that it becomes the new default? That topic seems to have
> gotten lost along the way somehow.

There is a Cygwin maintainer for git, yes.

-- 
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: Maintainer for git?

[Eric Blake]

[-- Attachment #1: Type: text/plain, Size: 875 bytes --]

On 02/11/2014 12:45 PM, Larry Hall (Cygwin) wrote:
> On 2/11/2014 2:03 PM, David Boyce wrote:
>> On Tue, Feb 11, 2014 at 9:58 AM, Adam Dinwoodie wrote:
>>>> Just curious where we stand with this now that 1.7.28 is out.
>>
>> Assuming that further testing affirms that the SSE2 bug affecting Git
>> is fixed in recent snapshots, this brings us back the original
>> subject: is there a maintainer for Git who will package up the latest
>> release such that it becomes the new default? That topic seems to have
>> gotten lost along the way somehow.
> 
> There is a Cygwin maintainer for git, yes.

Yes, but we are in the process of swapping who that maintainer is; the
next time git packages are pushed will be when the new maintainer is
ready to go.

-- 
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: Maintainer for git?

[Balaji Venkataraman]

Sorry it took me a while to try out all the cases.

On Wed, Jan 22, 2014 at 2:42 PM, Chris O'Bryan wrote:

> Could you try rerunning that same clone again with 1.8.5.2, but first
> temporarily replace cygcrypto-1.0.0.dll with a copy of
> cygcrypto-0.9.8.dll?

I tried that. It doesn't work at all - I just says 'Cloning into
'linux-git-1.8.5.2'...' and then returns to the prompt in < 2 seconds.
Nothing was cloned and no errors.

But what was really odd, is that when I went to copy 0.9.8 to 1.0.0, I
found two versions of cygcrypto-1.0.0.dll. One called just that and
another called cygcrypto-1.0.0.dll.new. I'm quite sure I have never
messed around with those files. So I moved the older (timestamp)
1.0.0.dll to .old and copied the .new version over as  and re-tried
the clone cygcrypto-1.0.0.dll. It failed in the usual fashion (pack
corrupted etc.) as reported earlier.

Wanting to eliminate any cygcrypto dll issues, I moved the existing
dlls out of the way and re-installed libopenssl. The dll that got
installed was identical in size and timestamp to .new but had
different md5sums! I've included all md5sums in case one of you can
verify which one you have. But the clone failed again.

All output below.

[This was w/ the cygcrypto-1.0.0.dll.new that was already present in /bin]
x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
linux-git-1.8.5.2
Cloning into 'linux-git-1.8.5.2'...
remote: Counting objects: 3366700, done.
remote: Compressing objects: 100% (507909/507909), done.
remote: Total 3366700 (delta 2833548), reused 3362709 (delta 2829741)
Receiving objects: 100% (3366700/3366700), 705.04 MiB | 3.02 MiB/s, done.
fatal: pack is corrupted (SHA1 mismatch)
fatal: index-pack failed

The last try (below) is after re-installing libopenssl100 via setup.exe.

x86$ cygcheck -f /bin/cygcrypto-1.0.0.dll
libopenssl100-1.0.1f-1

x86$ git clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
linux-git-1.8.5.2
Cloning into 'linux-git-1.8.5.2'...
remote: Counting objects: 3369475, done.
remote: Compressing objects: 100% (508418/508418), done.
remote: Total 3369475 (delta 2835968), reused 3365451 (delta 2832007)
Receiving objects: 100% (3369475/3369475), 704.55 MiB | 1.40 MiB/s, done.
Resolving deltas: 100% (2835968/2835968), done.
<------------- that step took hours to complete - never noticed that
before.
fatal: object of unexpected type
fatal: index-pack failed

x86$ \ls -l /bin/cygcrypto*
-rwxr-xr-x 1 <edit> Domain Users 1276957 Feb 12  2013 /bin/cygcrypto-0.9.8.dll
-rwxr-xr-x 1 <edit> Domain Users 1942016 Jan  8 07:25
/bin/cygcrypto-1.0.0.dll             <--- obtained from re-installing
libopenssl
-rwxr-xr-x 1 <edit> Domain Users 1942016 Jan  8 07:25
/bin/cygcrypto-1.0.0.dll.new      <--- was already present on the
machine
-rwxr-xr-x 1 <edit> Domain Users 1553920 Feb 12  2013
/bin/cygcrypto-1.0.0.dll.old

x86$ md5sum /bin/cygcrypto*
90cbe982fa8c13ae2be94e5c9a80a1ab */bin/cygcrypto-0.9.8.dll
09637556d4f11be4a89f92f15e988f9b */bin/cygcrypto-1.0.0.dll
<--- Note different md5sum from the file below.
d43f8ee26cce3c5925f17c6718ebe276 */bin/cygcrypto-1.0.0.dll.new
eab13dc5853adda51f681f74415418a5 */bin/cygcrypto-1.0.0.dll.old

--
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: Maintainer for git?

[Marco Atzeri]

On 23/01/2014 17:27, Balaji Venkataraman wrote:
> Sorry it took me a while to try out all the cases.
>

> Wanting to eliminate any cygcrypto dll issues, I moved the existing
> dlls out of the way and re-installed libopenssl. The dll that got
> installed was identical in size and timestamp to .new but had
> different md5sums! I've included all md5sums in case one of you can
> verify which one you have. But the clone failed again.
>

rebase changes the loading address so the md5sum will never match

Regards
Marco


--
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: Maintainer for git?

[Balaji Venkataraman]

On Thu, Jan 23, 2014 at 8:31 AM, Marco Atzeri wrote:
> On 23/01/2014 17:27, Balaji Venkataraman wrote:
>>
>> Sorry it took me a while to try out all the cases.
>>
>
>> Wanting to eliminate any cygcrypto dll issues, I moved the existing
>> dlls out of the way and re-installed libopenssl. The dll that got
>> installed was identical in size and timestamp to .new but had
>> different md5sums! I've included all md5sums in case one of you can
>> verify which one you have. But the clone failed again.
>>
>
> rebase changes the loading address so the md5sum will never match

I wasn't aware of that - good to know. 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: Maintainer for git?

[Andrey Repin]

Greetings, Balaji Venkataraman!

>>> Sorry it took me a while to try out all the cases.
>>>
>>
>>> Wanting to eliminate any cygcrypto dll issues, I moved the existing
>>> dlls out of the way and re-installed libopenssl. The dll that got
>>> installed was identical in size and timestamp to .new but had
>>> different md5sums! I've included all md5sums in case one of you can
>>> verify which one you have. But the clone failed again.
>>>
>>
>> rebase changes the loading address so the md5sum will never match

> I wasn't aware of that - good to know. Thanks.

That said, you probably need to run rebaseall. Worth a try.


--
WBR,
Andrey Repin (anrdaemon@yandex.ru) 23.01.2014, <21:46>

Sorry for my terrible english...


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

Re: Maintainer for git?

[Balaji Venkataraman]

On Thu, Jan 23, 2014 at 9:46 AM, Andrey Repin wrote:
> Greetings, Balaji Venkataraman!
>
>>>> Sorry it took me a while to try out all the cases.
>>>>
>>>
>>>> Wanting to eliminate any cygcrypto dll issues, I moved the existing
>>>> dlls out of the way and re-installed libopenssl. The dll that got
>>>> installed was identical in size and timestamp to .new but had
>>>> different md5sums! I've included all md5sums in case one of you can
>>>> verify which one you have. But the clone failed again.
>>>>
>>>
>>> rebase changes the loading address so the md5sum will never match
>
>> I wasn't aware of that - good to know. Thanks.
>
> That said, you probably need to run rebaseall. Worth a try.

Rebase didn't help. Still broken, IMO. I'm going back to 1.7.9 or will
rebuild w/ cygcrypto-0.9.8 as the OP had suggested.

x86$ git clone https://github.com/FFmpeg/FFmpeg.git ffmpeg
Cloning into 'ffmpeg'...
remote: Reusing existing pack: 345748, done.
remote: Counting objects: 613, done.
remote: Compressing objects: 100% (325/325), done.
remote: Total 346361 (delta 386), reused 484 (delta 288)
Receiving objects: 100% (346361/346361), 102.25 MiB | 2.93 MiB/s, done.
Resolving deltas: 100% (270961/270961), done.
fatal: object of unexpected type
fatal: index-pack failed

--
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: Maintainer for git?

[Charles Wilson]

On 1/23/2014 6:01 PM, Balaji Venkataraman wrote:

> Rebase didn't help. Still broken, IMO. I'm going back to 1.7.9 or will
> rebuild w/ cygcrypto-0.9.8 as the OP had suggested.
>
> x86$ git clone https://github.com/FFmpeg/FFmpeg.git ffmpeg
> Cloning into 'ffmpeg'...
> remote: Reusing existing pack: 345748, done.
> remote: Counting objects: 613, done.
> remote: Compressing objects: 100% (325/325), done.
> remote: Total 346361 (delta 386), reused 484 (delta 288)
> Receiving objects: 100% (346361/346361), 102.25 MiB | 2.93 MiB/s, done.
> Resolving deltas: 100% (270961/270961), done.
> fatal: object of unexpected type
> fatal: index-pack failed
>

I've found that I need to do the following:

PATH=/usr/bin git ....

or I get odd failures.  So I've actually set that up as an alias:

alias git='PATH=/usr/bin git'

--
Chuck


--
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: Maintainer for git?

[Charles Wilson]

On 1/28/2014 12:00 PM, Charles Wilson wrote:
> I've found that I need to do the following:
>
> PATH=/usr/bin git ....
>
> or I get odd failures.  So I've actually set that up as an alias:
>
> alias git='PATH=/usr/bin git'

NM. Should have read thru to the end of the thread before posting. Sorry 
for the noise.

--
Chuck



--
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: Maintainer for git?

[Andrey Repin]

Greetings, Charles Wilson!

>> Rebase didn't help. Still broken, IMO. I'm going back to 1.7.9 or will
>> rebuild w/ cygcrypto-0.9.8 as the OP had suggested.
>>
>> x86$ git clone https://github.com/FFmpeg/FFmpeg.git ffmpeg
>> Cloning into 'ffmpeg'...
>> remote: Reusing existing pack: 345748, done.
>> remote: Counting objects: 613, done.
>> remote: Compressing objects: 100% (325/325), done.
>> remote: Total 346361 (delta 386), reused 484 (delta 288)
>> Receiving objects: 100% (346361/346361), 102.25 MiB | 2.93 MiB/s, done.
>> Resolving deltas: 100% (270961/270961), done.
>> fatal: object of unexpected type
>> fatal: index-pack failed
>>

> I've found that I need to do the following:

> PATH=/usr/bin git ....

> or I get odd failures.  So I've actually set that up as an alias:

> alias git='PATH=/usr/bin git'

This sounds like you have issues with the default $PATH order.


--
WBR,
Andrey Repin (anrdaemon@yandex.ru) 29.01.2014, <01:00>

Sorry for my terrible english...


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

Re: Maintainer for git?

[Charles Wilson]

On 1/28/2014 4:02 PM, Andrey Repin wrote:
>> I've found that I need to do the following:
>
>> PATH=/usr/bin git ....
>
>> or I get odd failures.  So I've actually set that up as an alias:
>
>> alias git='PATH=/usr/bin git'
>
> This sounds like you have issues with the default $PATH order.

Nope, /usr/bin is in the front of my regular $PATH. Check the archives 
-- it has something to do with the size of the environment block being 
too big when spawning the new process. IIRC.  So making the $PATH short 
makes the env block smaller...

--
Chuck


--
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: Maintainer for git?

[Christopher Faylor]

On Wed, Jan 29, 2014 at 08:58:35AM -0500, Charles Wilson wrote:
>On 1/28/2014 4:02 PM, Andrey Repin wrote:
>>> I've found that I need to do the following:
>>
>>> PATH=/usr/bin git ....
>>
>>> or I get odd failures.  So I've actually set that up as an alias:
>>
>>> alias git='PATH=/usr/bin git'
>>
>> This sounds like you have issues with the default $PATH order.
>
>Nope, /usr/bin is in the front of my regular $PATH. Check the archives 
>-- it has something to do with the size of the environment block being 
>too big when spawning the new process. IIRC.  So making the $PATH short 
>makes the env block smaller...

Yes, environment size is a persistent, irritating problem.

cgf

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

Re: Maintainer for git?

[Steven Penny]

On Wed, Jan 15, 2014 at 4:55 AM, Adam Dinwoodie wrote
> This should now be fixed, with a new version uploaded to the same place.

Also for anyone interested in testing this is the install, pretty simple

    wget tastycake.net/~adam/cygwin/x86/git/git-1.8.5.2-1.tar.xz
    tar -x -C / -f git-1.8.5.2-1.tar.xz

--
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: Maintainer for git?

[Alexander Kriegisch]

I am wondering if this version will ever make it into the official distribution and be auto-updated when I run setup*.exe. Is there any progress? Just wondering, not demanding anything...
-- 
Alexander Kriegisch


> Am 21.01.2014 um 00:29 schrieb Steven Penny <svnpenn@gmail.com>:
> 
> On Wed, Jan 15, 2014 at 4:55 AM, Adam Dinwoodie wrote
>> This should now be fixed, with a new version uploaded to the same place.
> 
> Also for anyone interested in testing this is the install, pretty simple
> 
>    wget tastycake.net/~adam/cygwin/x86/git/git-1.8.5.2-1.tar.xz
>    tar -x -C / -f git-1.8.5.2-1.tar.xz
> 
> --
> 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: Maintainer for git?

[Adam Dinwoodie]

On Sat, Feb 15, 2014 at 09:11:44AM +0100, Alexander Kriegisch wrote:
> I am wondering if this version will ever make it into the official
> distribution and be auto-updated when I run setup*.exe. Is there any
> progress? Just wondering, not demanding anything...

As the (hopefully) incoming maintaner: yes, I'm very much hoping this
will become the version provided by the setup*.exes.  There are still a
few checks I want to run in the name of being sure I'm not making
anything worse, and some other details to iron out, but I am making
progress.

If you (or anyone else) wants to help speed things along, there're a few
tests you can do; I'm planning on doing these but if someone else can
it'll save me the time:

- Install git-cvs and the assorted dependencies mentioned in its
  setup.hint, and verify you can clone the Cygwin CVS repository.  I've
  not managed to do this without hitting errors, but I suspect that's
  because I'm using the tool incorrectly.

- Install minimal packages (Base + the dependencies listed in the
  setup.hint files) and check that the basic functionality works.

These all need to be done on both x86 and x86_64.  Since this is still
an unsupported package, to avoid list spam, please let me know off-list.
I'll summarize anything relevant to the list as appropriate.

(Incidentally, Alexander: please don't include other folks' email
addresses in your replies.  This mailing list is archived online and we
don't like feeding spam harvesters.)

Adam

--
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: Maintainer for git?

[Adam Dinwoodie]

On Sat, Feb 15, 2014 at 10:14:25AM +0000, Adam Dinwoodie wrote:
> On Sat, Feb 15, 2014 at 09:11:44AM +0100, Alexander Kriegisch wrote:
> > I am wondering if this version will ever make it into the official
> > distribution and be auto-updated when I run setup*.exe. Is there any
> > progress? Just wondering, not demanding anything...
> 
> As the (hopefully) incoming maintaner: yes, I'm very much hoping this
> will become the version provided by the setup*.exes.  There are still a
> few checks I want to run in the name of being sure I'm not making
> anything worse, and some other details to iron out, but I am making
> progress.
> 
> If you (or anyone else) wants to help speed things along, there're a few
> tests you can do; I'm planning on doing these but if someone else can
> it'll save me the time:
> 
> - Install git-cvs and the assorted dependencies mentioned in its
>   setup.hint, and verify you can clone the Cygwin CVS repository.  I've
>   not managed to do this without hitting errors, but I suspect that's
>   because I'm using the tool incorrectly.

I've tried this.  I have `git cvsimport` seemingly working on the
current Git 1.7.9 build, while my build reports the following SHA1
error:

~~~~
$ CVSROOT=:pserver:anoncvs@cygwin.com:/cvs/src git cvsimport -C cygwin -r cvs -k cygwin
Initialized empty Git repository in /home/Adam/vcs/cygwin/.git/
fatal: refs/remotes/cvs/master: not a valid SHA1
fatal: master: not a valid SHA1
fatal: You are on a branch yet to be born
checkout failed: 32768
~~~~

That's a showstopper, currently, and it'll probably take a little while
to work out where it's going wrong, since I have very little familiarity
with Git's internals.  I am working on it, but I don't have an ETA at
present.

(Of course, if anyone fancies helping out, or having a wonderful
revelation akin to Corinna's earlier OpenSSL one, that would be
exceedingly useful right now.)

> - Install minimal packages (Base + the dependencies listed in the
>   setup.hint files) and check that the basic functionality works.

I've done this, and everything seems to be fine there, at least for the
exceedingly basic tests I've done.

--
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: Maintainer for git?

[Achim Gratz]

Adam Dinwoodie writes:
> I've tried this.  I have `git cvsimport` seemingly working on the
> current Git 1.7.9 build, while my build reports the following SHA1
> error:
>
> ~~~~
> $ CVSROOT=:pserver:anoncvs@cygwin.com:/cvs/src git cvsimport -C cygwin -r cvs -k cygwin
> Initialized empty Git repository in /home/Adam/vcs/cygwin/.git/
> fatal: refs/remotes/cvs/master: not a valid SHA1
> fatal: master: not a valid SHA1
> fatal: You are on a branch yet to be born
> checkout failed: 32768
> ~~~~

I don't think that this error you see is related, but you absolutely
need cvsps version 2 for cvsimport to work correctly (or at all).


Regards,
Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+

Factory and User Sound Singles for Waldorf rackAttack:
http://Synth.Stromeko.net/Downloads.html#WaldorfSounds


--
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: Maintainer for git?

[Frank Fesevur]

2014-02-21 1:18 GMT+01:00 Adam Dinwoodie:
>> - Install git-cvs and the assorted dependencies mentioned in its
>>   setup.hint, and verify you can clone the Cygwin CVS repository.  I've
>>   not managed to do this without hitting errors, but I suspect that's
>>   because I'm using the tool incorrectly.
>
> I've tried this.  I have `git cvsimport` seemingly working on the
> current Git 1.7.9 build, while my build reports the following SHA1
> error:

Probably not going to solve this issue (I can't anything about CVS in
the changelog), but you noticed that 1.9.0 had been released recently?

Regards,
Frank

--
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: Maintainer for git?

[David Conrad]

On Wed, Feb 26, 2014 at 4:05 AM, Frank Fesevur wrote:
> 2014-02-21 1:18 GMT+01:00 Adam Dinwoodie:
>>> - Install git-cvs and the assorted dependencies mentioned in its
>>>   setup.hint, and verify you can clone the Cygwin CVS repository.  I've
>>>   not managed to do this without hitting errors, but I suspect that's
>>>   because I'm using the tool incorrectly.
>>
>> I've tried this.  I have `git cvsimport` seemingly working on the
>> current Git 1.7.9 build, while my build reports the following SHA1
>> error:
>
> Probably not going to solve this issue (I can't anything about CVS in
> the changelog), but you noticed that 1.9.0 had been released recently?
>
> Regards,
> Frank
>

Has there been any movement at all on git in the last six weeks?

What is "git-cvs"? There's no cygwin package by that name listed by setup.exe.

--
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: Maintainer for git?

[Adam Dinwoodie]

On Mon, Apr 14, 2014 at 03:40:17PM -0400, David Conrad wrote:
> On Wed, Feb 26, 2014 at 4:05 AM, Frank Fesevur wrote:
> > 2014-02-21 1:18 GMT+01:00 Adam Dinwoodie:
> >>> - Install git-cvs and the assorted dependencies mentioned in its
> >>>   setup.hint, and verify you can clone the Cygwin CVS repository.  I've
> >>>   not managed to do this without hitting errors, but I suspect that's
> >>>   because I'm using the tool incorrectly.
> >>
> >> I've tried this.  I have `git cvsimport` seemingly working on the
> >> current Git 1.7.9 build, while my build reports the following SHA1
> >> error:
> >
> > Probably not going to solve this issue (I can't anything about CVS in
> > the changelog), but you noticed that 1.9.0 had been released recently?
> >
> > Regards,
> > Frank
> >
> 
> Has there been any movement at all on git in the last six weeks?

Afraid not.  This is simply blocked on me finding time to do this, or
someone else somehow working out what's going wrong.

The problem appears to be something in my toolchain.  As I've previously
mentioned, the currently available version works correctly for me, but
if I try to build the same code, I get the same erroneous behaviour that
I see when I try to build a more recent version.

> What is "git-cvs"? There's no cygwin package by that name listed by setup.exe.

It's the part of Git that interacts with CVS repositories.  Notably,
this includes the Cygwin source code.  I've followed Yaakov's lead over
at Cygwin Ports in splitting this off from the main Git package, which
is why it's not (currently) a package in Cygwin.

--
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