Re: 2.9.5 and 2.1.73

[Sebastien Loisel]

David, what's would the difficulty level of producing a statically linked MinGW octave be? If that fixes the error() and cerr problems, I'd happily give up oct files in exchange of a working octave, at least to get started. I mean, I really like oct files, but I like a working octave even more.

I'm asking because I have a sneaking suspicion that the problem has something to do with libstdc++ being linked multiple times in DLLs and such, since libstdc++ is statically linked.

Sébastien Loisel

On 3/14/06, David Bateman <address@hidden> wrote:

> 5.0.2 installer comes with 3.4.4, so it wouldn't fix it.

Then you're saying that we need gcc 4.1 or newer? I don't want to have
to compile gcc for MinGW just for octave but rather use the current
stable version in MinGW. This leaves recompiling gcc 3.4.4 with the
right flag to avoid this bug as described in the bug report. Apparently
there is binary compatibility between the two versions of the compiler
is this case, but god what a mess. A better solution anyone? In fact
before setting out on such a horrid solution, can we confirm somehow
that this gcc bug is in fact the cause of our problem. If I understand
correctly then its a DLL issue, and so a fully static build of octave
shouldn't have this problem. You won't be able to use oct-files, so
maybe you won't be able to generate the problem in any case.

D.

--
David Bateman                                 address@hidden
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph)
Parc Les Algorithmes, Commune de St Aubin    +33 6 72 01 06 33 (Mob)
91193 Gif-Sur-Yvette FRANCE                  +33 1 69 35 77 01 (Fax)

The information contained in this communication has been classified as:

[x] General Business Information
[ ] Motorola Internal Use Only
[ ] Motorola Confidential Proprietary