Re: Octave Forge: requesting SVN access

[Carnë Draug]

On 5 October 2014 10:51, Oliver Heimlich <address@hidden> wrote:
> Am 04.10.2014 23:24, schrieb Carnë Draug:
>> On 2 October 2014 07:17, Oliver Heimlich <address@hidden> wrote:
>>> The package shall implement the upcoming IEEE 1788 standard on interval
>>> arithmetic. I am a member of the P1788 working group and the standard
>>> currently enters the sponsor ballot phase.
>>>
>>> My GNU octave package is in an early stage atm, but I could already
>>> submit some functionality to the public repository under GPL.
>>
>> I can create a repository where you can push to, would you prefer
>> mercurial or git?  In Octave, we have a preference for mercurial.
>
> Please create a mercurial repository then. Thanks.

Done [1, 2]. You should also push access to it, let me know if you
have any issue.
Also, we would appreciate if you could follow the Octave coding and commit
message guidelines [3].


>> On 2 October 2014 19:51, Oliver Heimlich <address@hidden> wrote:
>>> Please find attached a current development version of the code. A lot of
>>> functionality is still missing (I have been working on this during spare
>>> time for only a few days by now). Feedback is highly appreciated,
>>> because this is my first project in GNU Octave.
>>>
>>> The package comprises only M-files and depends on the package fenv.
>>
>> There is a problem with this because the fenv package has been
>> unmaintained for too long and is no longer an Octave Forge package.
>> Yes, we still have the tarball (there was only one release) available
>> for download on the archives, but it does not appear listed [1] and pkg's
>> "-forge" flag will no longer install it.
>>
>> I just gave the package a try and builds fine.  If it had a test suite
>> we could at least know if it's still working properly but as it is, I'd
>> guess no since its release was in 2008 and Octave has changed a lot since
>> then.
>>
>> Carnë
>>
>> [1] http://octave.sourceforge.net/packages.php
>
> The fenv package is only required for changing the processor's rounding
> mode for floating point arithmetics. I can certainly move this
> functionality into the interval package. Then it will be a standalone
> package.

If the fenv package does something useful to you, you could also bring it up
to life. If it is working, then I'd guess it would be enough to had some %!test
blocks to it, so at least we will when it really breaks.

Carnë

[1] https://sourceforge.net/p/octave/interval/ref/default/
[2] http://hg.code.sf.net/p/octave/interval
[3] http://wiki.octave.org/Commit_message_guidelines