TeamForge : Post

Forum Topic - Port of devel/scmgit-base for 6.4.0:

View: as

André Wösten(deleted)

06/01/2009 1:58 PM

post30446

Port of devel/scmgit-base for 6.4.0

Hello,

the next port I have done is GIT. Currently I'm working on porting X.org + XPhoton and I needed GIT for the modular X.
org enviroment.

Please commit the enclosed patch to the pkgsrc repository! Thanks!

Regards,
André

Attachment:

devel_scmgit-base.diff 1.72 KB

Sean Boudreau(deleted)

06/02/2009 6:27 PM

post30637

Re: Port of devel/scmgit-base for 6.4.0

On Mon, Jun 01, 2009 at 01:58:33PM -0400, André Wösten wrote:
> Hello,
> 
> the next port I have done is GIT. Currently I'm working on porting X.org + XPhoton and I needed GIT for the modular X.
org enviroment.
> 
> Please commit the enclosed patch to the pkgsrc repository! Thanks!

Yes, I came across this in git before and decided to
cogitate on it.  This patch doesn't give me a warm feeling.
Have you verified git will handle EINTR everywhere when
there's no SA_RESTART?  You might be able to send the
SIGALRM to a separate thread as in our ftp:

http://community.qnx.com/integration/viewcvs/viewcvs.cgi/trunk/utils/f/ftp/?root=core_networking&system=exsy1001

BTW the 'ed' issue mksh is complaining about will
be fixed in 6.4.2.  Here's a new diff you can send
them

Regards,

-seanb

Attachment:

patch-aa 411 bytes

André Wösten(deleted)

06/03/2009 4:24 AM

post30657

Re: Port of devel/scmgit-base for 6.4.0

Yes, you are right about git. Even if I've tested it intensively (it's working) - but to be safe, I will add some checks
 for EINTR in routines where for instance fgets is used.

I will post the patch in the next days. Thanks for the advice!

André

Sean Boudreau(deleted)

06/03/2009 8:56 AM

post30669

Re: Port of devel/scmgit-base for 6.4.0

On Wed, Jun 03, 2009 at 04:24:22AM -0400, André Wösten wrote:
> Yes, you are right about git. Even if I've tested it intensively (it's working) - but to be safe, I will add some 
checks for EINTR in routines where for instance fgets is used.
> 
> I will post the patch in the next days. Thanks for the advice!

To keep the diffs more self contained I'd personally
try the SIGALRM to a dedicated thread method.  The
git maintainers may be more inclined to accept them
in that case...

Regards,

-seanb

André Wösten(deleted)

06/07/2009 4:36 PM

post31054

Re: Port of devel/scmgit-base for 6.4.0

Sean, I worked on GIT. Enclosed the DIFF file - please take a look and tell me what you think about it.

BTW, the compiler warns about implicit declaration of sys/neutrino.h functions (SignalWaitinfo etc.). To remove the 
warnings,  I included sys/neutrino.h but I got an error about missing types in the itimer-structure, which is normally 
declared in sys/platform.h. But it never gets declared, because __EXT_QNX doesn't seem to be set correctly.

But probably I'm also missing something :-) I compile without including the QNX headers and link implicit. It's also 
working.

Regards,
André

Attachment:

devel_scmgit-base-2.diff 5.35 KB

Sean Boudreau(deleted)

06/15/2009 9:27 AM

post31722

Re: Port of devel/scmgit-base for 6.4.0

On Sun, Jun 07, 2009 at 04:36:42PM -0400, André Wösten wrote:
> Sean, I worked on GIT. Enclosed the DIFF file - please take a look and tell me what you think about it.
> 
> BTW, the compiler warns about implicit declaration of sys/neutrino.h functions (SignalWaitinfo etc.). To remove the 
warnings,  I included sys/neutrino.h but I got an error about missing types in the itimer-structure, which is normally 
declared in sys/platform.h. But it never gets declared, because __EXT_QNX doesn't seem to be set correctly.
> 
> But probably I'm also missing something :-) I compile without including the QNX headers and link implicit. It's also 
working.
> 
> Regards,
> André

Hi André:

Here's what I ended up with.  I think it's a little more self contained
(really only touches their Makefile).  I also cleaned up a few warnings.
Seems to work OK.  I checked this in on the HEAD_641 branch.

Regards,

-seanb

Attachment:

diff 6.74 KB

Return