On Fri, Jun 4, 2010 at 11:28 AM, Ævar Arnfjörð Bjarmason
<avarab@xxxxxxxxx> wrote:
> On Thu, Jun 3, 2010 at 21:22, Dario Rodriguez <soft.d4rio@xxxxxxxxx> wrote:
>> On Thu, Jun 3, 2010 at 5:21 PM, Ævar Arnfjörð Bjarmason
>>> What do you mean by "disappears" anyway, was it like $(echo|less), or
>>> did it just return with no output? What was the exit code?
>>>
>>
>> 'dissapears'? I will paste my output as is, there are 2 commits, but
>> 'git log' simply don't show them if $PAGER is not present:
>>
>> $ ../git log
>> $ echo $?
>> 0
>> $ PAGER=/bin/cat ../git log
>> commit 3274a12f940680612e3bfd3d022a0eab460c0f1f
>> Author: usuario tuxedo ####### <tx#####@MachineName.(none)>
>> Date: Thu Jun 3 20:02:23 2010 +0200
>>
>> OtherCom
>>
>> commit acf110f7c878a37e4a5af8499134df28da0e8ab3
>> Author: usuario tuxedo ####### <tx#####@MachineName.(none)>
>> Date: Thu Jun 3 20:01:37 2010 +0200
>>
>> inicial
>
> That's interesting.
>
Yes, I will check it better today at home.
>>> In any case, running git's make test might reveal other problems on
>>> AIX worth fixing. Maybe do that and post the results?
>>>
>>
>> The make test execution output is fairly long... do I post it all, or
>> attach MIME? However I'm leaving here and I cannot access the server
>> until tomorrow...
>
> The best thing would be to post cd t && ./$some_test -d -v for all
> tests that fail, I guess.
>
>>>>>> $ /usr/linux/bin/make prefix=$HOME/apps/ NO_OPENSSL=1 NO_TCLTK=1
>>>>>> NO_EXPAT=1 PYTHON_PATH=/usr/local/bin/python install
>>>>>>
>>>>>> [...]
>>>>>> install -d -m 755 '/myhomedir/apps/bin'
>>>>>> getopt: illegal option -- d
>>>>>> Usage: install [-c dira] [-f dirb] [-i] [-m] [-M mode] [-O owner]
>>>>>> [-G group] [-S] [-n dirc] [-o] [-s] file [dirx ...]
>>>>>> make: *** [install] Error 2
>>>>>>
>>>>>> Now the installing process fails because of the AIX 'install' tool and
>>>>>> I wonder, can I patch/configure the installing process for AIX? May be
>>>>>> a set of utils for building in such systems would help some people.
>>>>>
>>>>> Does AIX's install have something equivalent to GNU install's -d? The
>>>>> -c and -f options look likely from that synopsis.
>>>>>
>>>>
>>>> I don't know since I just use this system for development and testing
>>>> (I'm debian user), but let me post the manpage info, for -c and -f:
>>>>
>>>> -c DirectoryA Installs a new command file in the DirectoryA variable only if
>>>> that file does not already exist there. If it finds a copy of File there, it
>>>> issues a message and exits without overwriting the file. This flag can be used
>>>> alone or with the -s, -M, -O, -G, or -S flag.
>>>>
>>>> -f DirectoryB Forces installation of File in DirectoryB whether or not File
>>>> already exists. If the file being installed does not already exist, the command
>>>> sets the permission code and owner of the new file to 755 and bin, respectively.
>>>> This flag can be used alone or with the -o,-s, -M, -O, -G, or -S flag.
>>>
>>> Looks like there's no equivalent to -d. FWIW perl uses a installperl
>>> script that also works on AIX. Maybe a similar fallback or default
>>> would make sense for Git.
>>>
>>
>> Yes, I installed 'top' on other AIX machine today, and it uses it's
>> own install script too... may be it's the best way for systems having
>> a poor 'install' tool.
>>
>>>>>> PD2: I don't know if AIX python path is always /usr/local/bin/python,
>>>>>> but I've seen that git Makefiles set /usr/local/bin/python for FreeBSD
>>>>>> only:
>>>>>>
>>>>>> git_remote_helpers/Makefile:
>>>>>> ifndef PYTHON_PATH
>>>>>> ifeq ($(uname_S),FreeBSD)
>>>>>> PYTHON_PATH = /usr/local/bin/python
>>>>>> else
>>>>>> PYTHON_PATH = /usr/bin/python
>>>>>> endif
>>>>>> endif
>>>>>
>>>>> That's presumably because Python is most likely installed via the
>>>>> ports system on FreeBSD which drops it in /usr/local. How did you
>>>>> install Python on AIX? Is it from some IBM package or another method
>>>>> that's the most common & standard way to do it on AIX?.
>>>>>
>>>>
>>>> Again, I don't know since I'm not the sysadmin. I just looked for
>>>> python and found it's in /usr/local/bin
>>>
>>> Does using /usr/bin/env python instead work?
>>>
>>> $ cat /tmp/py.py
>>> #!/usr/bin/env python
>>> print "hello"
>>> $ /tmp/py.py
>>> hello
>>>
>>
>> Yes, it works... at least the executable is found :P
>>
>> $ cat temp.py
>> #!/usr/bin/env python
>> print "hello"
>>
>> $ ./temp.py
>> Could not find platform dependent libraries <exec_prefix>
>> Consider setting $PYTHONHOME to <prefix>[:<exec_prefix>]
>> hello
>
> That's one reason why I think we should just use /usr/bin/env for
> perl/python instead of hardcoding it to /usr/bin/{perl,python}. It's
> more likely to work on systems like AIX.
>
I agree... it's off topic, but something like this is what I say with
'test' for python, and I think those 2 are the only files impacted:
>From e89f18c29c171207002d4718ea4b9d08c7efe1d7 Mon Sep 17 00:00:00 2001
From: Dario Rodriguez <soft.d4rio@xxxxxxxxx>
Date: Fri, 4 Jun 2010 00:28:56 -0300
Subject: [PATCH] Now Makefiles check for Python executable
Check for python executable instead of predicting based on the system
---
Makefile | 9 +++++++--
git_remote_helpers/Makefile | 12 +++++++-----
2 files changed, 14 insertions(+), 7 deletions(-)
diff --git a/Makefile b/Makefile
index d5d6565..29cf8ff 100644
--- a/Makefile
+++ b/Makefile
@@ -449,7 +449,13 @@ ifndef PERL_PATH
PERL_PATH = /usr/bin/perl
endif
ifndef PYTHON_PATH
- PYTHON_PATH = /usr/bin/python
+ ifeq ($(shell test -x /usr/bin/python && echo y), y)
+ PYTHON_PATH = /usr/bin/python
+ else
+ ifeq ($(shell test -x /usr/local/bin/python && echo y), y)
+ PYTHON_PATH = /usr/local/bin/python
+ endif
+ endif
endif
export PERL_PATH
@@ -875,7 +881,6 @@ ifeq ($(uname_S),FreeBSD)
NO_UINTMAX_T = YesPlease
NO_STRTOUMAX = YesPlease
endif
- PYTHON_PATH = /usr/local/bin/python
HAVE_PATHS_H = YesPlease
endif
ifeq ($(uname_S),OpenBSD)
diff --git a/git_remote_helpers/Makefile b/git_remote_helpers/Makefile
index 74b05dc..8b5f50c 100644
--- a/git_remote_helpers/Makefile
+++ b/git_remote_helpers/Makefile
@@ -7,11 +7,13 @@ pysetupfile:=setup.py
DESTDIR_SQ = $(subst ','\'',$(DESTDIR))
ifndef PYTHON_PATH
- ifeq ($(uname_S),FreeBSD)
- PYTHON_PATH = /usr/local/bin/python
- else
- PYTHON_PATH = /usr/bin/python
- endif
+ ifeq ($(shell test -x /usr/bin/python && echo y), y)
+ PYTHON_PATH = /usr/bin/python
+ else
+ ifeq ($(shell test -x /usr/local/bin/python && echo y), y)
+ PYTHON_PATH = /usr/local/bin/python
+ endif
+ endif
endif
ifndef prefix
prefix = $(HOME)
--
1.7.1
Then, the path is checked if ever it's empty (after this stuff), but
the case is that no matter the system, '/usr/bin' and '/usr/local/bin'
are the most common places... I think it's totally sane the needing of
PYTHON_PATH if it's not in those commonly used dirs.
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html