Re: mingw buils

[Paul Kienzle]

Paul,

I got a running octave without readline and some broken stuff
which I don't remember at the moment.

At the time Ctrl-C was never going to work because windows
reports it to another thread, so the only way to break the current
context is to poll for it (obviously cygwin found a workaround,
and I chased through the code for a while before deciding it
was more than I wanted to put into Octave).  This is moot
now since Octave has gone the same route as windows,
requiring a poll for error in every loop [it's a mixed blessing---now
Octave is more careful with resources but less responsive to
Ctrl-C], so Windows style error handling will be easier now.

IIRC, John applied all the patches I sent him, so it may be a
simple matter of recompiling.  I haven't got back to it since Octave
went the polling route.

- Paul K

On Aug 9, 2004, at 2:41 PM, Paul Thomas wrote:

> I notice that there was a series of postings an a Mingw build around 
> October
> 2002, after which I find nothing.  Was this ever brought to the stage 
> of a
> complete working octave?
>
> The reason that I ask is that the more recently reported problems with 
> a
> gcc-3.3 build under Cygwin seem to be associated with Cygwin.dll 
> itself.  I
> can confirm the recent report of considerable changes in the time 
> taken by
> new/delete between g++ and g++ -mno-cygwin.  I find a factor of 6 and 
> some
> 60% with malloc/erase.  Even stranger is the factor of 12 for a 
> std::vector
> constructor.
>
> Paul T