octave-maintainers
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Interested in TISEAN package for GSoC


From: Piotr Held
Subject: Re: Interested in TISEAN package for GSoC
Date: Fri, 10 Apr 2015 20:11:33 -0600



W dniu poniedziałek, 6 kwietnia 2015 Juan Pablo Carbajal <address@hidden> napisał(a):
On Mon, Apr 6, 2015 at 8:52 PM, Piotr Held <address@hidden> wrote:
>
>
> W dniu poniedziałek, 6 kwietnia 2015 Juan Pablo Carbajal
> <address@hidden> napisał(a):
>>
>> On Mon, Apr 6, 2015 at 8:23 PM, Juan Pablo Carbajal
>> <address@hidden> wrote:
>> > On Mon, Apr 6, 2015 at 6:30 PM, Piotr Held <address@hidden> wrote:
>> >>
>> >>
>> >> 2015-04-04 10:19 GMT-06:00 Juan Pablo Carbajal <address@hidden>:
>> >>>
>> >>> On Sat, Apr 4, 2015 at 5:57 PM, Juan Pablo Carbajal
>> >>> <address@hidden> wrote:
>> >>> > On Sat, Apr 4, 2015 at 3:34 AM, Andreas Weber <address@hidden>
>> >>> > wrote:
>> >>> >> Am 03.04.2015 um 05:04 schrieb Piotr Held:
>> >>> >>> 2015-04-02 5:25 GMT-06:00 Juan Pablo Carbajal
>> >>> >>> <address@hidden
>> >>> >>> <mailto:address@hidden>>:
>> >>> >>>     TISEAN is not maintained anymore, So inheriting bugs is just
>> >>> >>> not
>> >>> >>> good.
>> >>> >>>     If TISEAN gets into Octave Forge I will take over maintenance
>> >>> >>> with
>> >>> >>> the
>> >>> >>>     help of whoever is actively developing for it.
>> >>> >>>     Also, I expect that in the long run TISEAN will eventually
>> >>> >>> dissolve
>> >>> >>>     into control, signal and tsa.
>> >>> >>>
>> >>> >>>     I will continue with the proposed methodology.
>> >>> >>>
>> >>> >>>
>> >>> >>> Juan,
>> >>> >>>
>> >>> >>> I have tried to looked for functions in octave that correspond to
>> >>> >>> the
>> >>> >>> TISEAN programs. The results of my research I have put into a
>> >>> >>> table
>> >>> >>> that
>> >>> >>> is accessible at the following link:
>> >>> >>>
>> >>> >>>
>> >>> >>> http://wiki.octave.org/User:Josiah425:TISEAN_Package:Table_of_functions
>> >>> >>>
>> >>> >>
>> >>> > Thanks Piotr,
>> >>> >
>> >>> > This is a nice step forward. Note that there is a folder "rountines"
>> >>> > inside source_c that also needs to be checked. The tar.gz file can
>> >>> > help you there. But what you did is the most important part.
>> >>
>> >> Well , the issue with rountines (in both source_c and source_f) is that
>> >> they
>> >> are not programs. They are only functions and constants used by the
>> >> programs
>> >> in the TISEAN package, so as far as I understand, the only time you
>> >> need to
>> >> worry about them is when compiling the oct file.
>> >>
>> >>> comment about the table:
>> >>> - Use one column for each thing. So comments should become several
>> >>> columns: "Language of origin"(C, Fortran), "How to port"(m-file, port,
>> >>> re-implement, etc...) and the other for "Status" (ported or not)
>> >>
>> >> I have changed the table to reflect these suggestions
>> >>
>> >> I have also started color coding the ones that are done, for clarity
>> >> purposes.
>> >>
>> >> Also I was thinking, should we try to create a separate page on the
>> >> wiki
>> >> about porting the TISEAN package? Or is it okay that the discussion of
>> >> the
>> >> programs in the TISEAN package is done as a subpage of my user page.
>> >>
>> > We will do this eventually. When we are ready to start. Lets file the
>> > rough edges first..and lets see if you get paid.
>> >
>> >> Also, can I still edit my elaborated proposal page? Can I incorporate
>> >> the
>> >> new information about functions that do not need to be ported, or does
>> >> that
>> >> proposal page need to be locked--in?
>> > Add everything to your proposal and add comments in your melange page
>> > to link to the improvements.
>> >
>> >>
>> >> :)
>> >>
>> Piotr,
>>
>> Remember that everytime you re-implement or replace a TISEAN function
>> we need to provide a test to check that it is equivalent.
>> Maybe you want to check autocor and corr already.
>
> Should I use the test functions, or is it enough if I get some example data
> and run it through both programs/functions?
>
> :)

I would get some example data (TISEAN has plenty) record the output of
TISEAN and then compare it to octave's. The comparison can be a test
function, but since this is run only in at port time, I would work on
a separate folder "port-check" or something a like. The color code in
your table could be YELLOW ported&untested, GREEN ported&testedOK.

We will prepare demos and test, do not worry we will go over those too.
Juan,

It took me a while, but I did what you suggested.

:) 

reply via email to

[Prev in Thread] Current Thread [Next in Thread]