Re: wgnuplot log errors to a file

[Tim H. <hof...@hi...>]

> A log file might be a good things to have. But I wonder if it would be worth aiming 
> for a portable solution - although we do not have the problem on other platforms.
> (Some people would prefer to drop the windows console completly, I guess. IMHO
> we should just make it just a bit more user friendly: readline support, more drag'n drop,
> descent menu layout and a new button bar etc.)
It definitly could need it a bit polishing. :)
I don't think that one needs a portable solution. All other systems can 
redirect the error stream which is more natural there (would also be my 
favorite solution in windows, but this does not work). So I would leave 
it in the windows part, except someone really wants it for all systems.

> The implementation for Windows via "MyPrintf" etc. would just be fine. Maybe putting the 
> code in TextPutStr()  (wtext.c) would be even simpler - one has to check.
In the My* functions there is still the file handle available. So I can 
check if the text is addressed to stderr and log only errors. Put 
apparently everything is coming via stderr, even the information at 
startup. I checked via (f==stderr) like it's done in the isterm() macro. 
Does someone know if this is right? Maybe it's a windows problem because 
the streams do not really exist.

However since I call with a plotfile as parameter all the terminal parts 
like startup information and prompt aren't written anyway.

> Btw. I would suggest -l <filename> instead of -log <filename>
that's fine by me.

Tim

-- 
Tim Hoffmann            Raum 052
HISKP                   Tel: +49-228-73-2942
Universitaet Bonn       Fax: +49-228-73-2505
Nussallee 14-16         hof...@hi...
53115 Bonn              http://www.hiskp.uni-bonn.de/gruppen/hikari