On May 22, 2013, at 5:45 AM, Jakub Hrozek wrote:
> On Wed, May 22, 2013 at 02:35:46PM +0200, Lukas Slebodnik wrote:
>> On (22/05/13 08:05), Stephen Gallagher wrote:
>>> On 05/22/2013 04:09 AM, Lukas Slebodnik wrote:
>>>> On (22/05/13 09:16), Sumit Bose wrote:
>>>>> On Wed, May 22, 2013 at 08:24:08AM +0200, Lukas Slebodnik wrote:
>>>>>> ehlo,
>>>>>>
>>>>>> Recommended way to create SRPM is to run make
>>>>>> (prerelease-)srpm. But in previous case make file have to be
>>>>>> generated, therefore configure script should not fail. (all
>>>>>> sssd required dependencies have to be installed) Script
>>>>>> make_srpm.sh can be runned without running configure, script
>>>>>> can be runned only from git repository.
>>>>>>
>>>>>>
https://fedorahosted.org/sssd/ticket/1927
>>>>>>
>>>>>> Patch is attached.
>>>>>
>>>>> Hi Lukas,
>>>>>
>>>>> I haven't tested the script yet, nevertheless I have two
>>>>> comments. Please add a copyright notice and a licence.
>>>>> Additionally I think '--define
>>>> I added copyright.
>>>>
>>>>> _source_filedigest_algorithm=1' should be switch on by an
option.
>>>>> Iirc we added it because RHEL5 could not handle newer hash
>>>>> algorithms here. But on newer platforms we should use the
>>>>> platform defaults.
>>>> I removed line with _source_filedigest_algorithm, because sssd
>>>> master no longer compiles on rhel5, nightly builds for rhel5 are
>>>> done from sssd-1-9 branch. But if you think, that it isn't proper
>>>> solution I can add new switch for this.
>>>>
>>>> Thank you very much for comments.
>>>>
>>>> New patch attached.
>>>>
>>>
>>> Almost an ack :)
>>>
>>> I am attaching two patches, one that should be squashed in and another
>>> that can go in on its own since it's related.
>>>
>>>
>>> Patch 0001: Fix English grammar and remove one sed line.
>>> There's no reason to substitute autoreconf into the spec file, as
it's
>>> already in there. The resulting spec would be calling it twice, which
>>> is wasteful.
>> I don't know why I thought, that there wasn't autoreconf.
>> Honestly, I didn't look to sssd.spec.in first time.
>> Thank you.
>>
>> I squashed the first patch to my patch.
>>
>>>
>>> Patch 0002: Given the argument above regarding
>>> _source_filedigest_algorithm (which is correct), I added a patch to
>>> remove that from the Makefile.am as well.
>> I am attaching the second patch unchanged.
>>
>> LS
>
> Hi Kevin, as you requested this script to be added, can you check if it
> does what you expect it to?
Hi Jakub,
Thanks very much for working on this issue.
I found a script called release.sh but need more details on running it.
Here's what I've tried:
1) check out the master branch (as per usual)
git clone
git://git.fedorahosted.org/git/sssd.git sssd
2) run script to create source tarball:
cd sssd
$ bash -x scripts/release.sh
+ SAVED_PWD=/local/home/svrbld/20130522-031502/sssd
++ grep '\[VERSION_NUMBER], \[.*\]' version.m4
++ grep '[0-9]\+\.[0-9]\+\.[0-9]\+' -o
+ version=1.9.93
++ echo 1.9.93
++ tr . _
+ tag=1_9_93
+ trap 'cd /local/home/svrbld/20130522-031502/sssd; rm -rf sssd-1.9.93
sssd-1.9.93.tar' EXIT
+ git archive --format=tar --prefix=sssd-1.9.93/ sssd-1_9_93
fatal: Not a valid object name
+ '[' 128 -ne 0 ']'
+ echo 'Cannot perform git-archive, check if tag 1_9_93 is present in git tree'
Cannot perform git-archive, check if tag 1_9_93 is present in git tree
+ exit 1
+ cd /local/home/svrbld/20130522-031502/sssd
+ rm -rf sssd-1.9.93 sssd-1.9.93.tar
Right, sorry, I wasn't specific enough. That script is currently under
review and I wanted to make sure it works for you before I push it to
master. Because it's just a single script, I'll send it to you directly.
Sorry for the confusion.