[ome-users] Makefile out-of-date with respect to Makefile.PL

[Ilya Goldberg]

I hope I'm not going to confuse the issue, but I think this is 
something that sometimes happens in hierarchical make files (make files 
that call other make files).
I don't understand why its happening in a clean checkout though.
Tomasz, why is there a Makefile copied at all?  Can't we just copy the 
matlab tree to /var/tmp/OME/install, then do everything from there - 
i.e. make, make test, make install?  This way there should never be a 
Makefile to copy - only a Makefile.PL
I'll be quiet now in case I've completely confused everyone.
-I

On Aug 5, 2005, at 2:24 PM, Anh Vu wrote:

> this is the content of /var/tmp/OME sometimes between installation and
> after it fails for the testing stage during installation.
>
> total 20
> drwxrwsr-x   4 avu ome 4096 Aug  5 11:06 Matlab
> drwxrwsr-x   3 ome ome 4096 Jul 23 00:43 install
> drwxrwsr-x  18 ome ome 4096 Jul 22 23:56 update
> drwxrwsr-x   2 ome ome 4096 Jul 22 23:19 lock
> drwxrwsr-x   2 ome ome 4096 Jul 22 23:19 sessions
> [avu at batch2112b OME]$ ls -lt
> total 16
> drwxrwsr-x   3 ome ome 4096 Jul 23 00:43 install
> drwxrwsr-x  18 ome ome 4096 Jul 22 23:56 update
> drwxrwsr-x   2 ome ome 4096 Jul 22 23:19 lock
> drwxrwsr-x   2 ome ome 4096 Jul 22 23:19 sessions
>
> and here is the content of /var/tmp/OME/Matlab for the brief moment
> that it is present during installation.
>
> total 108
> -rw-rw-r--  1 avu ome  3255 Aug  5 11:09 Makefile.PL
> -rw-rw-r--  1 avu ome  7089 Aug  5 11:09 MatlabInfo.pl
> -rw-rw-r--  1 avu ome   189 Aug  5 11:09 typemap
> -rw-rw-r--  1 avu ome   146 Aug  5 11:09 Changes
> drwxrwsr-x  2 avu ome  4096 Aug  5 11:09 CVS
> -rw-rw-r--  1 avu ome    57 Aug  5 11:09 MANIFEST
> -rw-rw-r--  1 avu ome 31827 Aug  5 11:09 Matlab.xs
> -rw-rw-r--  1 avu ome 26423 Aug  5 11:09 Makefile.old
> -rw-rw-r--  1 avu ome 14492 Aug  5 11:09 Matlab.pm
> -rw-rw-r--  1 avu ome  2578 Aug  5 11:09 test.pl
>
>
> On 8/5/05, T. J. Macura <tm289 at cam.ac.uk> wrote:
>> Dear Anh,
>>
>> As you can see below, the Vers/Arch/Home lines are transposed. There
>> was a transliteration problem that I fixed.
>> However this probably doesn't affect you because the lib and lib path
>> lines are correct.
>>
>>>  Matlab Vers: glnx86
>>>  Matlab Arch: /usr/local/matlab
>>>  Matlab Home: 7.0.4.352
>>>  Include:  -I/usr/local/matlab/extern/include
>>>  Lib:      -L/usr/local/matlab/bin/glnx86 -lmx -leng -lut -lmat
>>> -licudata -licui
>>>                                      18n -licuuc -lustdio -lz
>>>  Lib Path: /usr/local/matlab/bin/glnx86
>>
>> The way the Matlab-OME connector Test works is that
>> (src/perl2/OME/Matlab) is copied to a new temporary directory
>> (/var/tmp/OME/Matlab) whose ownership is set to the 'Matlab User'. 
>> This
>> copying step is necessary because the 'Matlab User' might not have
>> execute permissions in the folder where the OME distribution is
>> installed from.
>>
>> Then the perl script /src/perl2/OME/Matlab is run by the 'Matlab 
>> User'.
>> This script is run through make (i.e. make test) because make sets
>> various important environmental variables (e.g. LD_LIBRARY_PATH).
>>
>> I suspect that during the copying process on your machine, "Makefile"
>> is copied before "Makefile.PL". I added logic in the installer to
>> update 'Makefile' time-stamp to always be after Makefile.PL regardless
>> of the order they were copied over in.
>>
>> Please checkout from CVS and try it out,
>> If the Matlab Module Test fails again please do ls -lT for
>> /var/tmp/OME/Matlab
>>
>> Tom
>>
>>
> _______________________________________________
> ome-users mailing list
> ome-users at lists.openmicroscopy.org.uk
> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-users
>