new GSL version

new GSL version

[Patrick Alken]

Hi all,

   I think its time to make a new release of GSL (v1.16). There have 
been a number of bugfixes and new features added in the last couple 
years. I don't mind taking the lead on this but I was hoping to get a 
little feedback from others on whether there are any critical 
show-stopper bugs at the moment.

Yesterday I removed 3 tests from specfun, poly, and cdf which were 
causing 'make check' to fail on my system. I've created 3 separate bug 
reports to track these issues on savannah.

I admit I haven't been fully following all of the reported bugs, and I'm 
sure there are plenty on bug-gsl that never made it into the bug 
tracker. My thinking is that if there are no highly critical bugs at the 
moment, we should go ahead with the new release, and then focus on bug 
fixing for the following release. Many of the bugs in the tracker are 
outside my area of expertise and will involve significant effort to 
learn the algorithm to fix them, but I'm willing to spend some time in 
the coming months to look into as many of them as I can.

Also if any developers are current working on code they'd like to have 
in the next release, let me know that too and we can hold off for a 
little while.

Patrick

Re: new GSL version

[Brian Gladman]

On 24/05/2013 17:00, Patrick Alken wrote:
> Hi all,
> 
>   I think its time to make a new release of GSL (v1.16). There have been
> a number of bugfixes and new features added in the last couple years. I
> don't mind taking the lead on this but I was hoping to get a little
> feedback from others on whether there are any critical show-stopper bugs
> at the moment.
> 
> Yesterday I removed 3 tests from specfun, poly, and cdf which were
> causing 'make check' to fail on my system. I've created 3 separate bug
> reports to track these issues on savannah.
> 
> I admit I haven't been fully following all of the reported bugs, and I'm
> sure there are plenty on bug-gsl that never made it into the bug
> tracker. My thinking is that if there are no highly critical bugs at the
> moment, we should go ahead with the new release, and then focus on bug
> fixing for the following release. Many of the bugs in the tracker are
> outside my area of expertise and will involve significant effort to
> learn the algorithm to fix them, but I'm willing to spend some time in
> the coming months to look into as many of them as I can.
> 
> Also if any developers are current working on code they'd like to have
> in the next release, let me know that too and we can hold off for a
> little while.

Hi Patrick

I would certainly appreciate it if you are able to issue a new release
as I cannot access the GSL repository at all.

The problem is that I work on Windows where the current GSL repository
is unusable because of its use of symlinks which the Windows versions of
bzr do not implement.  So attempts to use the repository just report
errors.

As a result this means that I have not been able to build or test any of
the features that you and others have implemented since the last release.

I should also add that I have reported and offered fixes for some bugs,
at least one of which has not been acted on.  Although I realise that
Windows is not a priority for GSL, it would hardly make sense to leave
bugs in place just because they have been discovered and fixed on
Windows (in a fully portable way).

I should also say that I am willing to help out on GSL if the bzr
repository issue can be overcome in some way (there seems to be very is
little prospect that the Windows issue will be fixed since, as far as I
can tell, bzr development appears to have stopped).

   with my best regards,

     Brian Gladman

Re: new GSL version

[Patrick Alken]

I have a laptop running Mac OS X 10.6, and just ran make check on the 
latest repository with no errors. Does anyone have failures they can 
send to me?

On 05/24/2013 10:38 AM, Rhys Ulerich wrote:
>
> >   I think its time to make a new release of GSL (v1.16). There have 
> been a number of bugfixes and new features added in the last couple 
> years. I don't mind taking the lead on this...
>
> Thank you.
>
> I know of nothing critical but believe we want to have make check come 
> back clean at default optimization on OS X. It has not, I think due to 
> SVD tolerances, for some time. Do we have some sort of GNU-provided 
> continuous integration server available on which these types of 
> portability issues might be flushed out on an ongoing basis?
>
> I very much like your idea of just collecting up the bug reports and 
> moving on. Some sort of a post release bug fix effort would do well to 
> harness people's interest in contributing so that we might add some 
> pep to the community.
>
> - Rhys
>

Re: new GSL version

[Patrick Alken]

Hi Brian,

   I admit I haven't followed the windows issues closely at all.

   Is the symlink problem part of the GSL repository, or is it a 
windows/bzr issue? In my current repository, I see the following symlinks:

 > find . -type l
./siman/.gitignore
./doc/.gitignore
./.gitignore
./gsl/.gitignore

Is this what is causing you problems? I imagine all of these can be 
removed since we're not even using git anymore.

Regarding the windows bugs you've reported, are they in the bug tracker 
at all, or just floating around on the bug-gsl list? If they are not in 
the tracker, do you think you would be able to add them (or at least 
point me to the relevant bug-gsl dates), and if there are simple 
solutions, I could try fixing them early next week and sending you a 
test tarball file.

Patrick

On 05/24/2013 12:50 PM, Brian Gladman wrote:
> On 24/05/2013 17:00, Patrick Alken wrote:
>> Hi all,
>>
>>    I think its time to make a new release of GSL (v1.16). There have been
>> a number of bugfixes and new features added in the last couple years. I
>> don't mind taking the lead on this but I was hoping to get a little
>> feedback from others on whether there are any critical show-stopper bugs
>> at the moment.
>>
>> Yesterday I removed 3 tests from specfun, poly, and cdf which were
>> causing 'make check' to fail on my system. I've created 3 separate bug
>> reports to track these issues on savannah.
>>
>> I admit I haven't been fully following all of the reported bugs, and I'm
>> sure there are plenty on bug-gsl that never made it into the bug
>> tracker. My thinking is that if there are no highly critical bugs at the
>> moment, we should go ahead with the new release, and then focus on bug
>> fixing for the following release. Many of the bugs in the tracker are
>> outside my area of expertise and will involve significant effort to
>> learn the algorithm to fix them, but I'm willing to spend some time in
>> the coming months to look into as many of them as I can.
>>
>> Also if any developers are current working on code they'd like to have
>> in the next release, let me know that too and we can hold off for a
>> little while.
> Hi Patrick
>
> I would certainly appreciate it if you are able to issue a new release
> as I cannot access the GSL repository at all.
>
> The problem is that I work on Windows where the current GSL repository
> is unusable because of its use of symlinks which the Windows versions of
> bzr do not implement.  So attempts to use the repository just report
> errors.
>
> As a result this means that I have not been able to build or test any of
> the features that you and others have implemented since the last release.
>
> I should also add that I have reported and offered fixes for some bugs,
> at least one of which has not been acted on.  Although I realise that
> Windows is not a priority for GSL, it would hardly make sense to leave
> bugs in place just because they have been discovered and fixed on
> Windows (in a fully portable way).
>
> I should also say that I am willing to help out on GSL if the bzr
> repository issue can be overcome in some way (there seems to be very is
> little prospect that the Windows issue will be fixed since, as far as I
> can tell, bzr development appears to have stopped).
>
>     with my best regards,
>
>       Brian Gladman
>

Re: new GSL version

[Brian Gladman]

On 24/05/2013 21:50, Patrick Alken wrote:
> Hi Brian,
> 
>   I admit I haven't followed the windows issues closely at all.
> 
>   Is the symlink problem part of the GSL repository, or is it a
> windows/bzr issue? In my current repository, I see the following symlinks:
> 
>> find . -type l
> ./siman/.gitignore
> ./doc/.gitignore
> ./.gitignore
> ./gsl/.gitignore
> 
> Is this what is causing you problems? I imagine all of these can be
> removed since we're not even using git anymore.

I think it is but I won't know for certain until we try it.

If these symlinks could be removed, I will then see if the errors I am
seeing go away.

> Regarding the windows bugs you've reported, are they in the bug tracker
> at all, or just floating around on the bug-gsl list? If they are not in
> the tracker, do you think you would be able to add them (or at least
> point me to the relevant bug-gsl dates), and if there are simple
> solutions, I could try fixing them early next week and sending you a
> test tarball file.

I sent the details to Rhys some time ago but I can document the one I
fixed again.  I'll get back to you on this.

   Brian

Re: new GSL version

[Tuomo Keskitalo]

Patrick,

publishing 1.16 would be great! Awesome if you can push it out. I don't 
know of any active development.

BR,
Tuomo

On 05/24/2013 07:00 PM, Patrick Alken wrote:
> Hi all,
>
>    I think its time to make a new release of GSL (v1.16). There have
> been a number of bugfixes and new features added in the last couple
> years. I don't mind taking the lead on this but I was hoping to get a
> little feedback from others on whether there are any critical
> show-stopper bugs at the moment.
>
> Yesterday I removed 3 tests from specfun, poly, and cdf which were
> causing 'make check' to fail on my system. I've created 3 separate bug
> reports to track these issues on savannah.
>
> I admit I haven't been fully following all of the reported bugs, and I'm
> sure there are plenty on bug-gsl that never made it into the bug
> tracker. My thinking is that if there are no highly critical bugs at the
> moment, we should go ahead with the new release, and then focus on bug
> fixing for the following release. Many of the bugs in the tracker are
> outside my area of expertise and will involve significant effort to
> learn the algorithm to fix them, but I'm willing to spend some time in
> the coming months to look into as many of them as I can.
>
> Also if any developers are current working on code they'd like to have
> in the next release, let me know that too and we can hold off for a
> little while.
>
> Patrick
>


-- 
Tuomo.Keskitalo@iki.fi
http://iki.fi/tuomo.keskitalo