They are not deleted on my side (just tested) - SVN r9512
- Mail original -
De: "LAHAYE Olivier"
À: oscar-devel@lists.sourceforge.net
Envoyé: Jeudi 13 Décembre 2012 13:56:08
Objet: [Oscar-devel] opkgc: (make clean too clean?)
Doing the tests, I've noticed that issuing a
usr/share/doc/opkgc-1.0.1/src'
to directory
`/root/rpmbuild/BUILDROOT/opkgc-1.0.1-2.x86_64/usr/share/doc/opkgc-1.0.1/src'
cp: will not create hard link
`/root/rpmbuild/BUILDROOT/opkgc-1.0.1-2.x86_64/usr/share/doc/opkgc-1.0.1/templates'
to directory
`/root/rpmbuild/BUILDROOT/opkgc-
access
to SVN, feel free to check in.
- Mail original -
De: "LAHAYE Olivier"
À: oscar-devel@lists.sourceforge.net
Cc: valle...@ornl.gov
Envoyé: Jeudi 13 Décembre 2012 09:15:16
Objet: [Oscar-devel] opkgc SPEC Clean fix.
Geoffroy,
I've seen that you worked on opk
Geoffroy,
I've seen that you worked on opkgc spec file by removing man pages.
On CentOS-6.3" it realy works despite your comment. (same for de %doc section).
Anyway, here is a full fix of all issues in the spec file
1/ Manage the the manuals using wildcars so it works even on latest mandriva
l
Hello,
The patch is in (r9471).
Thanks for your contribution! :)
- Mail original -
De: "LAHAYE Olivier"
À: oscar-devel@lists.sourceforge.net
Envoyé: Mercredi 12 Septembre 2012 09:48:20
Objet: [Oscar-devel] opkgc patch: allow to overwrite destination if exists.
This patch a
This patch avoid oscar-packager to fail if destination already contains the
generated rpm (same name). Upon oscar-packager failure, the destination
/tftpboot/oscar/%ARCH% must be erased before retrying the build.
This can be a problem when some packages have been manualy generated (using
opkgc,
Spec file fixes (mainly doc and man paths that are hardcoded and wrong under
RHEL at least).
Best regards.
--
Olivier LAHAYE
CEA DRT/LIST/DCSI/DIR
Index: src/OpkgcLib/Rpm.py
===
--- src/OpkgcLib/Rpm.py (révision 9457)
+++ src
It is in (r9457). Thanks you so much for the contribution, that patch will
really help! OPKGC does not have any known bug now.
- Mail original -
De: "LAHAYE Olivier"
À: oscar-devel@lists.sourceforge.net
Envoyé: Lundi 6 Août 2012 09:27:43
Objet: [Oscar-devel] Opkgc Full fix
Hi,
Here is a fix to opkgc-0.6.1 so if fully works under Centos 6.3 (and CentOS-5.x
as well of course).
What does it addresses?
1/ Includes a needed patch not yet in svn: (see
http://www.mail-archive.com/oscar-devel@lists.sourceforge.net/msg09108.html for
the FIX and
http://www.mail-archive.
Hello,
It seems that maui and several other important packages are no longer
supported, even in the "unstable" release. Does anyone know the last
release that supported mpich and maui
Thanks
--
_
Hi,
In order to be able to run:
opkgc --dist=rhel --input=networking
from the ~svn/oscar/packages
I had to patch OpkgDescription.py
I think (not sure) that the 2 lines before the patch needs to be fixed as well.
Though, as arch is empty, the for loop isn't triggered, thus I can't tell if
its ok
Hi Jean,
If you have time, can you please look into this issue?
The problematic scripts are server-post-install and client-post-install.
The compilation of the corresponding opkg is just fine and the BUILD directory
has the files but the built rpm does not have the above scripts.
The odd thing
Hi,
Le 06.12.2007 18:24, Bernard Li a écrit :
> Hi Mike:
>
> On 12/6/07, Michael Edwards <[EMAIL PROTECTED]> wrote:
>
>
>> Has anyone gotten opkgc to compile on CentOS 4 (or RHEL4)?
>>
>> It seems to want python >= 2.4 which is bad because RHEL4 comes with 2.3.4.
>>
>> I am using CentOS 4 and I
Hi Mike,
I have setup python-2.4 on my RHEL4 machine several times. Notice that I did
not overwrite the existing python-2.3 by default on RHEL4 but put python-2.4 on
a different path (e.g., /opt/python-2.4) instead.
I believe that it won't make a trouble though. I know that it is a little
anno
Hi Mike:
On 12/6/07, Michael Edwards <[EMAIL PROTECTED]> wrote:
> Has anyone gotten opkgc to compile on CentOS 4 (or RHEL4)?
>
> It seems to want python >= 2.4 which is bad because RHEL4 comes with 2.3.4.
>
> I am using CentOS 4 and I tried to use the centosplus repo to update
> it but it couldn'
Has anyone gotten opkgc to compile on CentOS 4 (or RHEL4)?
It seems to want python >= 2.4 which is bad because RHEL4 comes with 2.3.4.
I am using CentOS 4 and I tried to use the centosplus repo to update
it but it couldn't find any updates.
I could update it by hand I guess, but then I expect mo
Hi Bernard,
Cool! Thank you very much. I have a cluster going on fc8 using Eric
Focht instructions and i had already uploaded some rpms i had built for
the platform.
I added two quad core nodes two days ago using oscar_wizard -manage and
it worked like a charm!
Thanks to all of you, your cre
Hi Jean:
On 12/2/07, Bernard Li <[EMAIL PROTECTED]> wrote:
> > Debs are on the OSCAR apt repository
> > RPMs : I have no RPM-like machine currently. spec file is ready on svn
> > (http://svn.oscar.openclustergroup.org/svn/oscar/pkgsrc/opkgc/packages/rpm/trunk/).
> > Could someone build it ?
>
> I
Hi Jean:
On 11/13/07, Jean Parpaillon <[EMAIL PROTECTED]> wrote:
> Tarball is available on GForge:
> http://oscar.gforge.inria.fr/downloads/opkgc-0.4.tar.gz
I added the URL to our wiki so that people can find the latest release:
http://svn.oscar.openclustergroup.org/trac/oscar/wiki/opkg_opkgc
Jean,
from what I know, we can only put one kind of such info into RPMs: the
PACKAGER tag. Already now I believe that we merge the three types of authors
into one. This is useless information. I'd expect that the length of what you
can put into the PACKAGER tag is limited to something.
If you nee
DongInn Kim wrote:
> OK, Jean.
>
> I am fine with having the author information for uploaders. BTW, if the
> author is already in the maintainer or upstream list, he/she does not need to
> be in the uploader list in order to put some changelogs, right?
>
No. There is _one_ author's list. Each
OK, Jean.
I am fine with having the author information for uploaders. BTW, if the author
is already in the maintainer or upstream list, he/she does not need to be in
the uploader list in order to put some changelogs, right?
Otherwise, I have to think about having the uploader list in config.xml
Hi,
DongInn Kim wrote:
> Hi Jean,
>
> So, are you saying that we have to keep the history of the list of uploaders
> in config.xml in order to put their changelog to it?
> Can we just assume that the authors on the changelog are the uploaders? (in
> fact, it is, right?)
>
From my point of vi
Hi,
Erich Focht wrote:
> On Tuesday 20 November 2007 05:29, DongInn Kim wrote:
>
>> For the 2nd case, I don't think that it is a good idea to restrict the
>> author of the changelog to the only authors of the specific packages.
>>
>>
>
> I agree. This is annoying and should be reverted.
Hi Jean,
So, are you saying that we have to keep the history of the list of uploaders in
config.xml in order to put their changelog to it?
Can we just assume that the authors on the changelog are the uploaders? (in
fact, it is, right?)
What is the benefit of having another list for only uploader
Hi,
Le 20.11.2007 19:48, Bernard Li a écrit :
> On 11/20/07, Erich Focht <[EMAIL PROTECTED]> wrote:
>
>
>> I agree. This is annoying and should be reverted. I don't want to force
>> the maintainer to update packages. If someone else can update it and the
>> maintainer
>> is in vacation, why no
Thanks,
I changed it in svn r6578.
Regards,
Erich
On Tuesday 20 November 2007 19:48, Bernard Li wrote:
> On 11/20/07, Erich Focht <[EMAIL PROTECTED]> wrote:
>
> > I agree. This is annoying and should be reverted. I don't want to force
> > the maintainer to update packages. If someone else can u
On 11/20/07, Erich Focht <[EMAIL PROTECTED]> wrote:
> I agree. This is annoying and should be reverted. I don't want to force
> the maintainer to update packages. If someone else can update it and the
> maintainer
> is in vacation, why not?
I'll put in my 2 cents worth of agreement with DongInn
On Tuesday 20 November 2007 05:29, DongInn Kim wrote:
> For the 2nd case, I don't think that it is a good idea to restrict the
> author of the changelog to the only authors of the specific packages.
>
I agree. This is annoying and should be reverted. I don't want to force
the maintainer to updat
Hi Jean,
After looking into the opkgc codes more carefully, I figured out the problems.
1. missing opkg.xsd was my fault because I did not fully regenerate the
"configure" file via autogen.sh once opkgc is up.
After regenerating the "configure" file via autogen.sh, getting the correct
/etc/op
Hi Jean,
Can you please take a look at opkgc:trunk to see what happened on the new opkgc?
> [INFO] Read config file /etc/opkgc.conf
> [ERROR] [Errno 2] No such file or directory: '@datadir@/@PACKAGE@/opkg.xsd'
This error happened when I had the following /etc/opkgc.conf.
> [EMAIL PROTECTED] packa
Hi all,
Le 11.11.2007 18:12, DongInn Kim a écrit :
> Hi Jean,
>
> Can you please build an opkgc rpm for the OSCAR users who want to try to test
> with the crispy branch?
> I found that it is a little difficult to compile opkgc and to make it work
> because the svn working copy of opkgc does not
Hi all,
I've just released a new version of opkgc: the 0.4. There are a lot of
changes and fixes.
Thanks to all who contributed.
Major changes are:
- manage distro filter with version with RPMs
- generate SRPMS
- add opkg-level dependencies:
- Requires, Provides and Conflicts are copied to
Hi Jean,
Can you please take a look at what is going on with compiling the "networking"
package with opkgc on YDL5-ppc64(ps3)?
I think it works fine on RHEL5-{X86,X86-64} and SuSE but I have no idea why it
keeps failing on YDL5-ppc64.
> [EMAIL PROTECTED] networking]# opkgc -v -v -v --dist ydl .
Hi Jean,
Can you please build an opkgc rpm for the OSCAR users who want to try to test
with the crispy branch?
I found that it is a little difficult to compile opkgc and to make it work
because the svn working copy of opkgc does not simply work after its
compilation. (i.e., I had to install Che
Le 05.11.2007 15:33, Erich Focht a écrit :
> Hi Jean,
>
Hi,
> I hope everything's fine...(?)
>
Sure, fresh and rested (I meant critical for OSCAR)
> We're just having trouble with the repositories at gforge.
> After the epoch number was deleted, the old packages are still considered the
Hi Jean,
I hope everything's fine...(?)
We're just having trouble with the repositories at gforge.
After the epoch number was deleted, the old packages are still considered the
good
ones (those with epoch included in version number).
Also: we need a restructuring of the yum repositories and the
Hi all,
I'm back after 1 critical month ;-) I hope everybody's fine.
Version numbers with "epoch" are allowed but not mandatories. With
automatic converter, if an epoch was there it has been kept but, of
course, version numbers can be simply 2.0 or even a single number.
Jean
Le 29.10.2007 11:5
Hi,
nobody cried, so I'd suggest we remove them from the config.xml files. Will
start today.
Regards,
Erich
On Friday 26 October 2007 03:52, DongInn Kim wrote:
> Hi Erich,
>
> That's what I was curious about it too. For me, the version number in front
> of the colon does not mean too much eith
Hi Erich,
That's what I was curious about it too. For me, the version number in front of
the colon does not mean too much either.
Regards,
- DongInn
Erich Focht wrote:
> A quick question: the automatic config.xml converter has created version
> numbers
> like 1:1.2.3-7 . For RPMs we usually
A quick question: the automatic config.xml converter has created version numbers
like 1:1.2.3-7 . For RPMs we usually get in addition to the number before the
":" the epoch.
Can't we simply drop the number in front of the colon? It seems to have no
particular use. I.e. let's replace the version nu
So i assume we do not try to "synchronize" Debian packages and RPM for this kind
of stuff. Because as i said before, i did do something like that for Debian few
weeks ago and i do not use the same names and i am not sure to split packages
for same way.
Selon Erich Focht <[EMAIL PROTECTED]>:
> Ber
Did you create a ticket?
Selon Erich Focht <[EMAIL PROTECTED]>:
> Hi,
>
> the config.xml offers filters for the distribution version. Therefore
> I think opkgc should also offer an option to specify a distribution version.
> There are differences between rhel-4 and rhel-5...
>
> Could this be add
> oscar-base : all OSCAR without packages/ and without prereqs packages.
I agree.
> oscar-base-client : makes no sense for me. but we need a dummy rpm
Libs only needed for client (no GUI stuff).
> oscar-base-server : same as above.
Libs and GUI stuff.
Selon Erich Focht <[EMAIL PROTECTED]>:
Hi Bernard,
my last checkins should fix the issue. "make baserpms" now builds also
oscar-base-server and oscar-base-client, so there is no need to modify opkgc
and rpm is in sync with deb. Only on rpm side e create additionally -scripts
and -libs. Maybe you or DongInn could test... Anyway, the opk
Hi Erich:
On 10/4/07, Erich Focht <[EMAIL PROTECTED]> wrote:
> Please modify the opkgc to require oscar-base-scripts on the clients (only
> on rpm based builds, please, we don't want to break debian). The
> oscar-base-scripts package will require oscar-base-libs.
>
> oscar-base is requiring both
Hi Erich:
On 10/4/07, Erich Focht <[EMAIL PROTECTED]> wrote:
> > However, this only fixes the dependency issue partially as
> > opkg--client/server still do not require opkg- -- I
> > think they should, right?
>
> No, you're wrong. opkg- was formelry known as opkg-api. It contains
> the API scrip
Bernard,
I just made a spec file and changed the Makefile such that
% make baserpms
will build you three oscar-base rpms.
The RPMs will be:
oscar-base
oscar-base-libs
oscar-base-scripts
See:
https://svn.oscar.openclustergroup.org/trac/oscar/changeset/6241
The perl "provides" are generated aut
Hi,
the config.xml offers filters for the distribution version. Therefore
I think opkgc should also offer an option to specify a distribution version.
There are differences between rhel-4 and rhel-5...
Could this be added, please?
Thanks,
regards,
Erich
Hi Bernard,
forgot to mention again:
Rebuilding the opkg-* metapackages makes no sense now. All automatically
converted config.xml files are possibly broken. See the example below. They
need to be fixed manually. The issue is that all dependencies without a
filter specifying either oscar_client o
Hi Bernard,
On Wednesday 03 October 2007 03:29, Bernard Li wrote:
> http://svn.oscar.openclustergroup.org/trac/oscar/changeset/6235
yes, the issue seems solved. Thanks!
> opkgc needs to be rebuilt and all the opkg- RPMs will also
> need to be rebuilt and uploaded to gforge.
As said, the gforge
Hi Erich:
On 10/2/07, Erich Focht <[EMAIL PROTECTED]> wrote:
> The second (and more major) issue: The metaRPMs built by opkgc contain no
> requires at all! It is fine that they contain only a spec file, but it seems
> to be wrong.
>
> > rpm -q --requires -p opkg-sis-server-1:3.8.0-3.noarch.rpm
>
Hi,
I can confirm that some things are completely bogus right now.
Looking at packages/sis/config.xml : it looks much more unreadable than it ever
was. But worse: the clientDeps are empty while the apiDeps get some entries
which should be common to server and client. The apiDeps should never show
I have filed both issues in the bug tracker:
http://svn.oscar.openclustergroup.org/trac/oscar/ticket/431
http://svn.oscar.openclustergroup.org/trac/oscar/ticket/432
Cheers,
Bernard
On 10/2/07, DongInn Kim <[EMAIL PROTECTED]> wrote:
> FYI,
>
> Here is a good example to show what problem I have.
FYI,
Here is a good example to show what problem I have.
http://www.osl.iu.edu/~dikim/opkg-sis-1:3.8.0-3.src.rpm
This srpm was generated while opkgc generated the SIS opkg meta rpms but looks
like it does not include everything that we need.
> [EMAIL PROTECTED] SRPMS]# rpm -qpl opkg-sis-1\:3.8.0
Hi Jean,
Can you please fix the dependency problems on opkg meta rpms ASAP? More
specifically
1. opkg--server rpm does not have the proper "requires".
> [EMAIL PROTECTED] noarch]# rpm -q --requires -p
> opkg-sis-server-1\:3.8.0-3.noarch.rpm
> /bin/sh
> /bin/sh
> rpmlib(CompressedFileNames)
Hi all,
I have a question for the community: how do we deal with switcher when using
OPKGC? For instance, the config.xml file for LAM has:
mpi
lam-7.1.2
How do we specify that with the new format? and more generally, how do we plan
to use switcher when we will have
Hi Jean,
There is an error while I am registering to the
[EMAIL PROTECTED] mailing list.
>
> Bug in Mailman version 2.1.5
>
>
> We're sorry, we hit a bug!
>
> Please inform the webmaster for this site of this problem. Printing of
> traceback and other system information has been explici
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Le 10.09.2007 20:52, DongInn Kim a écrit :
> Jean,
>
> I think my account (donginn) is not allowed to view/upload contents on
> projects/oscar page of https://gforge.inria.fr?
It is now. Don't forget to add your ssh public key to upload packges,
wait
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Le 10.09.2007 20:11, DongInn Kim a écrit :
> Hi Jean,
>
> Cool, it works fine on my RHEL5 now.
>
:-)
> So, what do I have to do now?
> Do we have any dumb and simple instruction of how an OSCAR developer can
> contribute the compiled opkgc?
Yes,
h
Hi Jean,
Cool, it works fine on my RHEL5 now.
So, what do I have to do now?
Do we have any dumb and simple instruction of how an OSCAR developer can
contribute the compiled opkgc?
Regards,
- DongInn
Jean Parpaillon wrote:
> Hi,
> Sorry for being laate
>
> Le 04.09.2007 18:37, DongInn Kim
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi,
Sorry for being laate
Le 04.09.2007 18:37, DongInn Kim a écrit :
> Hi Jean,
>
> Can you please look into the problem(#401) that I filed a few weeks ago?
> I don't think I can go further without this thing fixed.
> http://svn.oscar.opencluster
Hi Jean,
Can you please look into the problem(#401) that I filed a few weeks ago?
I don't think I can go further without this thing fixed.
http://svn.oscar.openclustergroup.org/trac/oscar/ticket/401
If you have any workaround, can you please point it out for me?
Regards,
--
- DongInn
---
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi,
Le 21.08.2007 08:21, Bernard Li a écrit :
> Hi all:
>
> On 8/20/07, DongInn Kim <[EMAIL PROTECTED]> wrote:
>
>> http://svn.oscar.openclustergroup.org/trac/oscar/wiki/opkgAPI
>
> Going through this document, I noticed a typo.
>
> "third-part" s
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi,
Thank you for testing opkgc ;-)
Le 21.08.2007 08:04, DongInn Kim a écrit :
> I installed opkgc-0.3.1-1 rpm from the link of Jean's repository that
> Bernard pointed out for me.
>
Having a look at oscar.gforge.inria.fr/downloads/, the latest tar
Hi all:
On 8/20/07, DongInn Kim <[EMAIL PROTECTED]> wrote:
> http://svn.oscar.openclustergroup.org/trac/oscar/wiki/opkgAPI
Going through this document, I noticed a typo.
"third-part" should be "third-party". It seems this is also the case
in the .xsd file:
http://svn.oscar.openclustergroup.or
I installed opkgc-0.3.1-1 rpm from the link of Jean's repository that
Bernard pointed out for me.
This opkgc seems to work fine but..
> [EMAIL PROTECTED] sis]# opkgc --dist=rhel
> [INFO] Read config file /etc/opkgc.conf
> [INFO] Validating XML file against schema
> Error: ./scripts/server-post-in
Hi Jean,
I have a few questions about opkgc.
1. I am using this one that you have posted on the devel list before.
http://oscar.gforge.inria.fr/downloads/opkgc-0.3.tar.gz
Am I working with the right opkgc?
Looks like that the man pages of opkgc are all empty. Is it correct?
> [EMAIL PROTECTED]
Hi,
I just checked-in a small script for the automatic compilation of OPKGs
included in the "Default" package set of your local distribution. Therefore,
if your favorite Linux distro is not supported (binary packages not available
via the OSCAR repositories), you can try to compile the OSCAR pa
Hi,
I've released a new version of opkgc which fix some bugs (see tickets)
and is a little bit more strict:
- opkg name ( tag in config.xml) must comply with RPM and Deb
package name: [a-z][a-z0-9+.-]*
- summary tag in config.xml is better formatted
- description tag is formatted in output files, a
About this point, i think the idea was to delete that script a long time
ago... but was never effectively done. At least this is what i remember from
a discussion with Thomas. :-)
My 2 cents,
Le lundi 16 juillet 2007 09:15, Jean Parpaillon a écrit :
> > post_install api-post-deploy
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi,
lxml is a library above libxml2. It is an implementation of an API
sometimes called ElementTree.
Googling lxml on OpenSuSe, I have this:
http://rpmfind.net/linux/RPM/opensuse/OSS-factory/i586/python-lxml-1.1.2-33.i586.html
Hope that can help.
Reg
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi all,
pre-note: this discussion is about the API of OSCAR packaging, it could
be *very* interesting that a lot of people give its opinion.
Then,
I agree that there are too much API calls but I think the most we use
deb and rpm facilities, the less
Hi Jean,
I cannot start opkgc on openSuse:
[EMAIL PROTECTED]:~/Projects/Cluster/oscar-svn/trunk> opkgc --help
Traceback (most recent call last):
File "/usr/local/bin/opkgc", line 28, in
from Opkgc.OpkgcCompiler import *
File "/usr/local/lib64/python2.5/site-packages/Opkgc/OpkgcCompiler.p
Hi Jean,
looking at
http://svn.oscar.openclustergroup.org/trac/oscar/wiki/opkgAPI
I get the impression that the number of API calls has doubled. This is not
good, the API may offer every thinkable possibility but is actually the
problem. People get confused and don't know which of the many choices
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi,
Le 13.07.2007 17:02, Erich Focht a écrit :
> On Friday 13 July 2007 15:46, Jean Parpaillon wrote:
>>> OTOH, the post_client_rpm_install script is actually meant to be run after
>>> the image was created, so this actually is the api-post-image. Thi
On Friday 13 July 2007 15:46, Jean Parpaillon wrote:
> > OTOH, the post_client_rpm_install script is actually meant to be run after
> > the image was created, so this actually is the api-post-image. This one
> > runs chrooted in the image, so doesn't have access to all info.
> >
>
> Huh, I'm not s
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Le 13.07.2007 12:09, Erich Focht a écrit :
> Jean,
>
> while converting the mpich package I wondered about one thing:
> The post_clients API script is renamed to api-post-image.
>
Note 1: old->new name renaming is based on informations available in
Jean,
while converting the mpich package I wondered about one thing:
The post_clients API script is renamed to api-post-image.
The post_clients script is supposed to run each time after new clients were
defined in the cluster. It is not supposed to run after the image was
created. I suggest we re
Michael, Erich, Geoffroy,
Thank you very much for the comments and info. I have decided (as suggested)
to build the package based on the old method as it is tried, tested and
documented. From what I gather it seems that converting old packages to the
new package system shouldn't be too hard. In t
Hi Stuart,
I second Erich and Michael: opkgc is still under development it is better to
use the old format. However, i have a question for the group: is the
documentation about OPKGs fixed? Last time i had a look at this documentation
it was kind of out-of-date (some script names were wrong and
Hi Stuart,
I agree with Michael, the best thing to do is to create a package in the
"old" format. That means you will need to write a config.xml, build RPMs
and put them into the distro/$DISTRO$VERSION-$ARCH subdirectory. Also put
the srpms to SRPMS/ .
The basic idea of the new opkg format is to
Well, one option is to make it in the current, better documented
package format and convert it to the new format with a script which
was written to convert the existing packages (also undocumented).
As the documentation guy, I apologize, but I have not gotten around to
looking at the documentation
Dear all,
I have just had a quick word with bernardl on IRC and he said to bring to
your attention that I feel that there are some documentation/clarity issues
to do with OPKGC. I am struggling and want people to know why so things can
be made better.
I have been plodding along with MPITB (MPI T
Hi all,
You can find a new version of opkgc with deb and RPM packages here:
http://oscar.gforge.inria.fr/downloads/
Regards,
Jean
--
Kerrighed inside - OSCAR outside
http://kerrighed.org/
Hi Jean,
my sed version is 4.1.5. If that helps...
Thanks, best regards,
Erich
On Wednesday 27 June 2007 15:25, Jean Parpaillon wrote:
> Hi,
>
> Le Tuesday 26 June 2007 11:30:56 Erich Focht, vous avez écrit :
> > Hi,
> >
> > ... closely related to the topic:
> >
> > I'm having trouble with the
not in response to Erich,
I also a compilation problem
1. At revision 5963.
2. [EMAIL PROTECTED] trunk]# make
3. Making all in src
4. make [1]: Entering directory `/opt/soc/pkgsrc/trunk/src'
5. python setup.py build
6. running build
7. running build_py
8. running build_scripts
Hi,
... closely related to the topic:
I'm having trouble with the "make install" step of opkgc (motivated by the
attempt to convert some packages. It says:
Making install in src
make[1]: Entering directory
`/home/focht/Projects/Cluster/oscar-svn/pkgsrc/opkgc/opkg_exp/src'
python setup.py build
Hi all,
Jean and I start today to convert all OPKGs to the new format (using opkgc).
That will most certainly break trunk for few days and piss off most of the
developers but we really need this feature for OSCAR 5.1 and therefore we
would like to make the transition ASAP.
If you have problems
Hi Jean,
This is a very good news! I have one question for you (i cannot find a quick
answer reading the code or the documentation): does the opkg- package
depends on other binary package related to the given OPKG?
I will try to finish OPD2 ASAP now that we have this point addressed.
Thanks,
Hi all,
I've just released opkgc 0.2 which add original config.xml in opkg-
packages, as required by (many ?) pieces of codes in OSCAR.
Once the package opkg- is installed, config.xml is available in
/var/lib/oscar/packages//config.xml
opkgc is available here:
http://oscar.gforge.inria.fr/dow
Oops, sorry. Here is a new install.log
Jean Parpaillon wrote:
> Hi DongInn,
>
> Le vendredi 20 avril 2007 19:01, Kim, DongInn a écrit :
>
>> Hi Jean,
>>
>>
>>> [EMAIL PROTECTED] opkgc-0.1]# man opkg
>>> No manual entry for opkg
>>> [EMAIL PROTECTED] opkgc-0.1]# man opkgc
>>> No manual entr
Hi DongInn,
Le vendredi 20 avril 2007 19:01, Kim, DongInn a écrit :
> Hi Jean,
>
> > [EMAIL PROTECTED] opkgc-0.1]# man opkg
> > No manual entry for opkg
> > [EMAIL PROTECTED] opkgc-0.1]# man opkgc
> > No manual entry for opkgc
> > [EMAIL PROTECTED] opkgc-0.1]# opkgc
> > Traceback (most recent call
Hi Jean,
> [EMAIL PROTECTED] opkgc-0.1]# man opkg
> No manual entry for opkg
> [EMAIL PROTECTED] opkgc-0.1]# man opkgc
> No manual entry for opkgc
> [EMAIL PROTECTED] opkgc-0.1]# opkgc
> Traceback (most recent call last):
> File "/usr/local/bin/opkgc", line 28, in ?
> from Opkgc.OpkgcCompile
Hi all,
Here it is. I've just released a v.0.1 version of opkgc. You can download it,
waiting for a better place:
http://oscar.gforge.inria.fr/downloads/opkgc-0.1.tar.gz
It includes:
- opkgc
- opkgc doc: supposed to explain how opkgc works, but it's not really
finished. Did not seem to me to be
Le jeudi 22 mars 2007 19:28, Steven Blackburn a écrit :
> If the new config files contain the same sort of data, would it be possible
> / efficient use of time to write an xslt file to convert old files to the
> new format?
Yes, that make sense.
Anyway, I have to write 2 xslt: either 1 for old and
Hi all,
> On Thursday 22 March 2007 14:34, Jean Parpaillon wrote:
>> I've commited a draft of a new config.xml schema.
>> I tried not to break everything but some things needs changing, in its
>> actual
>> form, most packages won't build with the new schema. Of course, this is
>> a
>> draft so it
On Sunday 25 March 2007 19:09, Geoffroy VALLEE wrote:
> On Saturday 24 March 2007 05:04, Erich Focht wrote:
> > This is unacceptable for me personally and is a BAD THING to do in an
> > open source project
>
> [...]
>
> > By the way, opkgc is another candidate for this kind of issues. To make it
Jean,
> >> My idea is:
> s/My idea is/An idea of Ericht Focht I totally agree with/
[...]
thanks, accepted.
> (and I spent some hours writting the implementation of your idea,
> operating system does not yet handle well pure ideas)
And I'm always happy to see ideas implemented.
> I have an ide
On Saturday 24 March 2007 05:04, Erich Focht wrote:
> This is unacceptable for me personally and is a BAD THING to do in an
> open source project
[...]
> By the way, opkgc is another candidate for this kind of issues. To make it
> clear: the opkg compiler (opkgc) is an idea developed by Brian Fin
1 - 100 of 104 matches
Mail list logo
