Bug#936924: Moving libsvm to Debian Science team

[Chen-Tse Tsai]

Hi Christian,

Yes, my changes are merged.

I won't be able to work on it in the next few days. Thanks for your help!
If you want, it would be great to add you as an Uploader since you know
these packages really well.

Thanks,
Chen-Tse


On Fri, Jan 10, 2020 at 8:14 AM Christian Kastner  wrote:

> On 10.01.20 11:07, Andreas Tille wrote:
> > since my upload was a bit questionable this gives us the chance
> > to discuss it again.  What option would you prefer:
> >
> >1. I just re-upload what is in Git right now to new queue
> >2. Somebody who feels more competent takes over
> >3. Something else
>
> I dropped the ball on this when I got back from the holidays, but I can
> look into it now.
>
> Chen-Tse, is your work already in the git repo on Salsa?
>

Bug#936924: Moving libsvm to Debian Science team

[Chen-Tse Tsai]

Got it. Thanks! I missed the c++filt part.

Chen-Tse


On Wed, Dec 25, 2019 at 12:27 PM Christian Kastner  wrote:

> On 2019-12-25 18:08, Chen-Tse Tsai wrote:
> > Thanks Andreas and Christian for the updates/comments. I agree that I
> > should have been more verbose in the changelog. I'm tracing all the
> > commits to learn more about packaging. I was playing with gbp for
> > updating upstream. I really like how it works with git! No worries, my
> > time was not wasted anyway.
>
> If you like gbp, then it's already a team win :-)
>
> >
> > I talked to Christian briefly about the symbol file. He added one for
> > liblinear4. I have a question about this file. The symbol file I
> > generated looks a bit different from liblinear4.symbols, in which
> > several lines start with (c++) and contain the exact parameters. I
> > couldn't figure out how to make dpkg-gensymbols produce this.
>
> Yeah, the C++ symbols generated are the mangled versions. You need to
> unmangle them by piping the generated file through c++filt, and then add
> the (c++) prefix to tell dpkg-gensymbols about this. The
> UsingSymbolsFiles [1] page has a practical example on the bottom.
>
> The c++filt(1) and dpkg-gensymbols(1) man pages have theoretical
> background on this, but TBH, I don't think it's needed.
>
> There are voices that say that symbols files for C++ libraries are
> overly hard to maintain, but in case of this specific upstream, I didn't
> have a negative experience, and it did help me detect breaking changes a
> few times.
>
> [1] https://wiki.debian.org/UsingSymbolsFiles
>

Bug#936924: Moving libsvm to Debian Science team

[Chen-Tse Tsai]

Thanks Andreas and Christian for the updates/comments. I agree that I
should have been more verbose in the changelog. I'm tracing all the commits
to learn more about packaging. I was playing with gbp for updating
upstream. I really like how it works with git! No worries, my time was not
wasted anyway.

I talked to Christian briefly about the symbol file. He added one for
liblinear4. I have a question about this file. The symbol file I generated
looks a bit different from liblinear4.symbols, in which several lines start
with (c++) and contain the exact parameters. I couldn't figure out how to
make dpkg-gensymbols produce this.

Thanks,
Chen-Tse


On Wed, Dec 25, 2019 at 10:02 AM Andreas Tille  wrote:

> On Wed, Dec 25, 2019 at 03:04:47PM +0100, Christian Kastner wrote:
> > On 2019-12-25 10:59, Andreas Tille wrote:
> > > Kind regards
> > >
> > >Andreas.
> >
> > I've re-read my original message and feel that I was overly harsh. I was
> > irritated about the rushed upgrade as in private communication with
> > Chen-Tse, I highlighted precisely the possible complexity of this
> > upgrade and how it should be performed properly -- but, that was private
> > communication.
>
> I admit I did not considered it overly harsh.  My upload was in fact not
> the best idea and your response just reflected it.
>
> > Apologies.
>
> Thanks but from my point of view not needed.  Everything is fine at my
> side and as I said I'll now leave the issue to you who are more
> comfortable with that package.  Something that came to my mind later:
> When upstream does not mark SOVERSIONs:  In this case I usually add a
> symbols file.  Thats more maintenance burden but usually helps to spot
> issues with ABI changes.
>
> Kind regards
>
>Andreas.
>
>
> --
> http://fam-tille.de
>

Bug#936924: Moving libsvm to Debian Science team

[Chen-Tse Tsai]

Thanks, Andreas.

I just submitted a PR for dropping python2 dependencies:
https://salsa.debian.org/science-team/libsvm/merge_requests/1

Any comment is appreciated. I'll be working on upgrading to new upstream.

Thanks,
Chen-Tse


On Mon, Dec 23, 2019 at 4:03 PM Andreas Tille  wrote:

> On Mon, Dec 23, 2019 at 03:06:37PM -0500, Chen-Tse Tsai wrote:
> > I have a quick question. Previously the python modules are installed to
> > /usr/share/pyshared/. Should I use /usr/lib/python3/dist-packages instead
> > if I change it to python3? I see this in the policy and this is also what
> > liblinear uses. Just want to double check.
>
> Yes, follow liblinear example.
>
> Thanks for your work on this
>
>  Andreas.
>
> --
> http://fam-tille.de
>

Bug#936924: Moving libsvm to Debian Science team

[Chen-Tse Tsai]

I have a quick question. Previously the python modules are installed to
/usr/share/pyshared/. Should I use /usr/lib/python3/dist-packages instead
if I change it to python3? I see this in the policy and this is also what
liblinear uses. Just want to double check.

Thanks,
Chen-Tse


On Mon, Dec 23, 2019 at 2:29 AM Andreas Tille  wrote:

> Hi Chen-Tse,
>
> On Sun, Dec 22, 2019 at 09:47:44PM -0500, Chen-Tse Tsai wrote:
> > I'm working on an update which removes dependency on cdbs and python2.
> I'm
> > using liblinear package as the reference. I'll send you something for
> > review some time this week.
>
> Thanks a lot.  That's really appreciated
>
>   Andreas.
>
> --
> http://fam-tille.de
>

Bug#936924: Moving libsvm to Debian Science team

[Chen-Tse Tsai]

Hi Christian, All,

I'm working on an update which removes dependency on cdbs and python2. I'm
using liblinear package as the reference. I'll send you something for
review some time this week.

Thanks,
Chen-Tse


On Sat, Dec 21, 2019 at 4:26 PM Christian Kastner  wrote:

> Hi all,
>
> Regarding liblinear: I thought I already set the Maintainer to Debian
> Science Team, I guess I missed it.
>
> On 2019-12-21 20:11, Chen-Tse Tsai wrote:
> > I've created an account on Salsa.
> >
> > Do you think we should remove cdbs and use another build system
> > instead? Please let me know if you have any suggestion. I'm not
> > familiar with other build systems.
>
> Yep, the Debian Policy was updated and recommends dh now.
>
> @Andreas: Chen-Tse and I discussed upgrading the package in October (I
> think), but both did not have the time back then.
>
> I could also help with some work starting on the 27th or so. As
> src:liblinear and src:libsvm have the same upstream, they are quite
> similar WRT to building, so src:liblinear (which is up to date) might
> have some helpful ideas.
>
> Given that the last libsvm update was a while ago, I wouldn't be
> surprised if a transition were necessary.
>

Bug#936924: Moving libsvm to Debian Science team

[Chen-Tse Tsai]

I've created an account on Salsa.

Do you think we should remove cdbs and use another build system instead?
Please let me know if you have any suggestion. I'm not familiar with other
build systems.

Thanks
Chen-Tse


On Sat, Dec 21, 2019 at 12:27 PM Andreas Tille  wrote:

> Hi Chen-Tse,
>
> thanks for you quick response.
>
> On Sat, Dec 21, 2019 at 09:48:14AM -0500, Chen-Tse Tsai wrote:
> > Hi, Andreas,
> >
> > I agree with your suggestions. Thanks for putting it on salsa!
>
> Thanks for confirming.
>
> > Should I investigate dropping python2? I can spend some time this week on
> > this. But I may need help with debian stuff.
>
> What exact help do you need?  Do you have a login on Salsa?  We could
> add you to science-team to grant you commit permissions.  I admit I have
> not yet looked into the packaging - just realised that its cdbs which is
> unfortunate since it would not eliminate the Python2 dependency fully.
>
> Kind regards
>
>   Andreas.
>
> > On Sat, Dec 21, 2019 at 7:39 AM Andreas Tille  wrote:
> >
> > > On Sat, Dec 21, 2019 at 08:28:57AM +, Mo Zhou wrote:
> > > > I second this proposal, and the same for src:liblinear.
> > >
> > > That's done as I wrote.
> > >
> > > > These are high popcon packages, dependencies for a number of other
> > > > packages. They should be team maintained to unblock important fixes.
> > >
> > > To push a bit I just commited
> > >
> > >https://salsa.debian.org/science-team/libsvm
> > >
> > > Last maintainer upload was more than 3 years ago, 2 NMUs inbetween,
> > > package is lagging behind upstream.  Anybody is kindly invited to adapt
> > > the packaging (I think we should really get rid of cdbs since this in
> > > turn is Python2) and proceed from here.
> > >
> > > Kind regards
> > >
> > >  Andreas.
> > >
> > >
> > > > On Sat, Dec 21, 2019 at 08:35:28AM +0100, Andreas Tille wrote:
> > > > > Hi Chen-Tse,
> > > > >
> > > > > I'm maintaining a package that depends from libsvm.  Due to bug
> #936924
> > > > > that did not received any response since August it is in danger to
> be
> > > > > removed from testing so I'm interested that this bug will be fixed.
> > > > > When looking at the package I realised that while it would fit into
> > > > > Debian Science team scope it is not team maintained nor is there
> any
> > > > > repository in Salsa.  That's why I have the following suggestion:
> > > > >
> > > > >   1. I create a repository on Salsa
> > > > >   2. Move the package to Debian Science team maintenance
> > > > >  and add you as Uploader
> > > > >   3. Drop Python2 package and close bug #936924
> > > > >   4. May be migrate packaging from cdbs to dh
> > > > >
> > > > > If I do not hear from you until Monday I assume you agree with this
> > > > > plan and will do so.
> > > > >
> > > > > Kind regards
> > > > >
> > > > >  Andreas.
> > > > >
> > > > > PS: @Christian: I noticed that you and Chen-Tse are maintaining
> > > > > liblinear.  I have just removed Python2 package and reassigned
> > > > > #936889 to ftp.debian.org to make sure python-liblinear will
> be
> > > > > removed.  Thus libsvm can be dealt as suggested.
> > > > >
> > > > > --
> > > > > http://fam-tille.de
> > > > >
> > > >
> > > >
> > >
> > > --
> > > http://fam-tille.de
> > >
>
> --
> http://fam-tille.de
>

Bug#936924: Moving libsvm to Debian Science team

[Chen-Tse Tsai]

Hi, Andreas,

I agree with your suggestions. Thanks for putting it on salsa!

Should I investigate dropping python2? I can spend some time this week on
this. But I may need help with debian stuff.

Thanks,
Chen-Tse


On Sat, Dec 21, 2019 at 7:39 AM Andreas Tille  wrote:

> On Sat, Dec 21, 2019 at 08:28:57AM +, Mo Zhou wrote:
> > I second this proposal, and the same for src:liblinear.
>
> That's done as I wrote.
>
> > These are high popcon packages, dependencies for a number of other
> > packages. They should be team maintained to unblock important fixes.
>
> To push a bit I just commited
>
>https://salsa.debian.org/science-team/libsvm
>
> Last maintainer upload was more than 3 years ago, 2 NMUs inbetween,
> package is lagging behind upstream.  Anybody is kindly invited to adapt
> the packaging (I think we should really get rid of cdbs since this in
> turn is Python2) and proceed from here.
>
> Kind regards
>
>  Andreas.
>
>
> > On Sat, Dec 21, 2019 at 08:35:28AM +0100, Andreas Tille wrote:
> > > Hi Chen-Tse,
> > >
> > > I'm maintaining a package that depends from libsvm.  Due to bug #936924
> > > that did not received any response since August it is in danger to be
> > > removed from testing so I'm interested that this bug will be fixed.
> > > When looking at the package I realised that while it would fit into
> > > Debian Science team scope it is not team maintained nor is there any
> > > repository in Salsa.  That's why I have the following suggestion:
> > >
> > >   1. I create a repository on Salsa
> > >   2. Move the package to Debian Science team maintenance
> > >  and add you as Uploader
> > >   3. Drop Python2 package and close bug #936924
> > >   4. May be migrate packaging from cdbs to dh
> > >
> > > If I do not hear from you until Monday I assume you agree with this
> > > plan and will do so.
> > >
> > > Kind regards
> > >
> > >  Andreas.
> > >
> > > PS: @Christian: I noticed that you and Chen-Tse are maintaining
> > > liblinear.  I have just removed Python2 package and reassigned
> > > #936889 to ftp.debian.org to make sure python-liblinear will be
> > > removed.  Thus libsvm can be dealt as suggested.
> > >
> > > --
> > > http://fam-tille.de
> > >
> >
> >
>
> --
> http://fam-tille.de
>

Bug#832075: libsvm: diff for NMU version 3.21+ds-1.1

[Chen-Tse Tsai]

Hi Olly,

Thanks for your help!

Chen-Tse

On Fri, Jul 29, 2016 at 10:29 PM, Olly Betts  wrote:

> Control: tags 832075 + patch
> Control: tags 832075 + pending
>
> Dear maintainer,
>
> I've prepared an NMU for libsvm (versioned as 3.21+ds-1.1) and
> since the fix is obvious and the effects very annoying for anyone with
> libsvm3 installed, I've uploaded it under the zero-day NMU rules:
>
> https://www.debian.org/doc/manuals/developers-reference/pkgs.html#nmu
>
> Cheers,
> Olly
>

Bug#716363: [Mayhem] Bug report on libsvm-tools: svm-scale crashes with exit status 139

[Chen-Tse Tsai]

Hi,

Thanks for reporting the bug. Since the link of your script is not
available now, I couldn't reproduce the bug. Could you please provide it
again?

Thanks,
Chen-Tse

Bug#804343: RFS: libsvm/3.20-1 [NMU] -- library implementing support vector machines

[Chen-Tse Tsai]

Hi,

Sorry for the late reply as I missed your message earlier. I'll update the
package as soon as possible. Thanks for the help!

Regards,
Chen-Tse

Bug#667791: [PATCH] libsvm: Helping to update to packaging format 3.0

[Chen-Tse Tsai]

Hi Jari,

Thanks for your notification. All the modification looks great. Please
upload it in a NMU.

Best Regards,
Chen-Tse


jari.aa...@cante.net 於 2012年4月6日下午11:43 寫道：

 Package: libsvm
 Severity: wishlist
 Tags: patch

 Hi,

 The dpatch patch management system has been deprecated for some time. The
 Lintian currently flags use of dpatch packages as an error. The new 3.0
 packaging format is an improved version which, among other things, contains
 patch management built-in. For more information, see:

http://wiki.debian.org/Projects/DebSrc3.0

 I had some free time; see attached patch to migrate to new package
 format. Note that all files in debian/patches/* are canocalized to
 *.patch.

 Let me know if there is anything that needs adjusting or if it is ok
 to upload this version in a NMU in case you are working on other
 issues needing attention.

 Thanks,
 Jari

Bug#608851: libsvm-dev: no need of versioned include/ subdirectory

[Chen-Tse Tsai]

Thanks for the report!
I'll fix it soon.

Best Regards,
Chen-Tse


2011/1/4 Yaroslav Halchenko deb...@onerussian.com

 Package: libsvm-dev
 Version: 3.0-1
 Severity: normal


 Since there is no libsvm3.0-dev, and multiple versions (e.g. 2.91 and 3.0)
 of
 libsvm-dev could not be co-installed on the system, it really makes little
 to
 no sense to carry additional versioned directory such as
 /usr/include/libsvm-3.0 on top of libsvm includes directory.  Or am I
 missing
 something?

 Such include directories migration from one location to another might for
 no
 good reason ruin builds of dependent packages which might remain
 compatible with multiple versions of libsvm while not foreseeing a variety
 of
 versioned directories coming with new major releases  of libsvm.  Partially
 it
 accounts for FTBFS of pymvpa (#608844, CCed).

 -- System Information:
 Debian Release: squeeze/sid
  APT prefers testing
  APT policy: (900, 'testing'), (800, 'unstable'), (300, 'experimental')
 Architecture: amd64 (x86_64)

 Kernel: Linux 2.6.32-5-amd64 (SMP w/2 CPU cores)
 Locale: LANG=en_US, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
 Shell: /bin/sh linked to /bin/bash

 Versions of packages libsvm-dev depends on:
 ii  libsvm3   3.0-1  library implementing support
 vecto

 libsvm-dev recommends no packages.

 libsvm-dev suggests no packages.

 -- no debconf information

Bug#557339: libsvm: FTBFS without Python 2.4

[Chen-Tse Tsai]

Thanks for the bug report.
I am working on this and other problems reported earlier.

Thanks,
Chen-Tse

Bug#557241: Please add Chen-Tse Tsai to debian-maintainers keyring

[Chen-Tse Tsai]

Package: debian-maintainers
Version: 1.64
Severity: wishlist

Hello,

I would like to be added to the debian-maintainers keyring.
See my declaration:

http://lists.debian.org/debian-newmaint/2009/11/msg00039.html

Thanks,

Chen-Tse


add-EA11D609624D95E5
Description: Binary data

Bug#543852: ncurses -- shared libraries for terminal handling

[Chen-Tse Tsai]

I'm interested in this package but still studying it.
If there is any team want to pick it, I would very like to join. :)

Chen-Tse

Bug#517408: marked as done (libsvm: libsvm reports 'nu' for C-SVM and vice-versa)

[Chen Tse Tsai]

Package: libsvm2
Version: 2.89-1

Chen-Tse

2009/4/24 Debian Bug Tracking System ow...@bugs.debian.org:

 Your message dated Fri, 24 Apr 2009 10:08:26 +0800
 with message-id 2df58ee30904231908l7285d538wef9cb3bb917b7...@mail.gmail.com
 and subject line bug fixed
 has caused the Debian Bug report #517408,
 regarding libsvm: libsvm reports 'nu' for C-SVM and vice-versa
 to be marked as done.

 This means that you claim that the problem has been dealt with.
 If this is not the case it is now your responsibility to reopen the
 Bug report if necessary, and/or fix the problem forthwith.

 (NB: If you are a system administrator and have no idea what this
 message is talking about, this may indicate a serious mail system
 misconfiguration somewhere. Please contact ow...@bugs.debian.org
 immediately.)


 --
 517408: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=517408
 Debian Bug Tracking System
 Contact ow...@bugs.debian.org with problems


 -- 轉寄訊息 --
 From: Michael Hanke michael.ha...@gmail.com
 To: Debian Bug Tracking System sub...@bugs.debian.org
 Date: Fri, 27 Feb 2009 15:08:05 +0100
 Subject: libsvm: libsvm reports 'nu' for C-SVM and vice-versa
 Package: libsvm2
 Version: 2.85.0-1
 Severity: normal
 File: libsvm

 Hi,

 in the progress report LIBSVM also includes the parameters of the
 running SVM. However, it reports a 'nu' parameter for C-SVM and a 'C'
 parameter for Nu-SVM.

 It looks like a simple typo in solve_nu_svc() and solve_c_svc() (defined
 in svm.cpp) respectively. This bug is also present in the current 2.88
 upstream release.

 Thanks,

 Michael



 -- System Information:
 Debian Release: 5.0
  APT prefers testing
  APT policy: (500, 'testing')
 Architecture: i386 (i686)

 Kernel: Linux 2.6.26-1-686 (SMP w/1 CPU core)
 Locale: LANG=de_DE.UTF-8, LC_CTYPE=de_DE.UTF-8 (charmap=UTF-8)
 Shell: /bin/sh linked to /bin/bash

 Versions of packages libsvm2 depends on:
 ii  libc6 2.7-18 GNU C Library: Shared libraries
 ii  libgcc1   1:4.3.3-3  GCC support library
 ii  libstdc++64.3.3-3The GNU Standard C++ Library v3

 libsvm2 recommends no packages.

 libsvm2 suggests no packages.

 -- no debconf information




 -- 轉寄訊息 --
 From: Chen Tse Tsai ctse.t...@gmail.com
 To: 517408-d...@bugs.debian.org
 Date: Fri, 24 Apr 2009 10:08:26 +0800
 Subject: bug fixed
 Package: libsvm2
 Version: 2.89-1

 Hi,

 This isn't a bug.
 Because the connection between C-SVM and nu-SVM,
 From C we can calculate nu and vise versa.

 Thanks,
 Chen-Tse






--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org