[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: octave-3.6.2 vs octave-4.0.0-rc4: sparsersb.oct: derived class of oc
From: |
Michele Martone |
Subject: |
Re: octave-3.6.2 vs octave-4.0.0-rc4: sparsersb.oct: derived class of octave_sparse_matrix in octave_value_list "ignored" |
Date: |
Tue, 19 May 2015 11:02:17 +0200 |
User-agent: |
Mutt/1.5.21 (2010-09-15) |
On address@hidden:54, Michele Martone wrote:
> Hi Octave Maintainers.
>
> My `sparsersb' package [s] used to build/work under Octave, e.g. 3.6.2.
> I gave a try to octave-4.0.0-rc4.
>
> The `sparsersb' keyword continues returning a octave_value_list object having
> been append()'ed (in position 0) to `retval'.
> "class octave_sparsersb_mtx : public octave_sparse_matrix"
>
> In turns out that under octave-3.6.2 what sparsersb() returns is
> being interpreted correctly (as octave_sparsersb_mtx).
>
> What is being returned by octave-4.0.0-rc4 is wrong: it appears
> as "Compressed Column Sparse" once it is being printed out.
> This, rather than "Recursive Sparse Blocks".
>
> So it's like something wrong in the innheritance mechanism.
>
> Can anybody point out what changed from octave-3.6.2 to
> octave-4.0.0 w.r.t. this problem and how to correct in
> sparsersb.cc ?
>
> [s]
> svn+ssh://address@hidden/p/octave/code/trunk/octave-forge/main/sparsersb/src
Ok, I fixed this using the now corrected "make_int.cc" example:
https://savannah.gnu.org/bugs/index.php?45136
The problem was with the different signature of
virtual bool octave_base_value::print(...)
in Octave 4.
Now sparsersb works with 3.6.2, 3.8.2, 4.0.0-rc4.
I'ld say it is mature enough for a package release (together with
librsb-1.2, released a few weeks ago).
Help in this would be very appreciated !