Re: 2.9.19??

[Tatsuro MATSUOKA]

Hello Michael

I also think that your way in the pgnuplot should be a unique solution to 
realise the 'true'
pipe-gnuplot in the literal sense.

I read the threads that you have indicated.  I also felt disappointed to the 
not so active attitude of
the gnuplot people.

Anyway I will go to gnuplot ML again and will propose again the replacement of 
current pgnuplot by you
use in the octave MSVC release. 
Perhaps my proposal will not be accepted considering my poor writing abilities.
However, I would like state that the current pgnuplot by Hans is obsolete and 
should be replaced.

Finally I will state really appreciation to you for your information and your 
efforts so far. 

I'll go to gnuplot ML from now! :-)

Regards


Tatsuro

--- Michael Goffioul <address@hidden> wrote:

> On 12/11/07, Tatsuro MATSUOKA <address@hidden> wrote:
> > > On 12/11/07, Tatsuro MATSUOKA <address@hidden> wrote:
> > > > Anyway Michael's gnuplot goes well, this is a trivial issue but I 
> > > > wonder why.
> > >
> > > Probably because the way pgnuplot and wgnuplot communicate together
> > > is not suited for binary data.
> >
> > OK!
> > I think that Michael's opnion is right.
> >
> > The original pgnuplot just send data
> > in a byte unit wgnuplot windows by Win32 API.
> >
> > This seems to the gnuplot problem.
> >
> > I'll go to the gnuplot ML.
> 
> That's basically to avoid this kind of problems (and the one you also
> tried to work around by adding a delay in pgnuplot.c) that I developed
> a real console mode for gnuplot, with real pipes (I don't find the existing
> pgnuplot solution very elegant).
> 
> I proposed a patch back in April this year (before the 4.2 release), but
> the follow up is not very active. See
> http://sourceforge.net/tracker/index.php?func=detail&aid=1627936&group_id=2055&atid=302055
> 
> Michael.
> 


--------------------------------------
New Design Yahoo! JAPAN  2008/01/01
http://pr.mail.yahoo.co.jp/newdesign/