Re: [LINUX5250] cvs updated

[Scott Klement <tn5250@xxxxxxxxxxxxxxxx>]

> I'm wondering about lib5250.so.  Because cursesterm.c is compiled as 
> part of the build, lib5250 depends on curses.  This shouldn't really be 
> necessary considering that all the curses code should be in the curses 
> terminal, not the 5250 library.  For example, x5250 has no need of 
> curses support.  I'd like to be able to build lib5250 without curses or 
> slang or any terminal at all.

It seems reasonable that each terminal should get it's own subdirectory, 
and should be built separately from lib5250.  I don't have time to do the 
work, though.

Maybe we should have a cursesterm subdirectory, an x5250 subdirectory... 
(there's already a win32 subdirectory) and put the sources applicable to 
each into those directories with their own Makefiles, etc.

We could control which of them gets built based on options passed to the 
configure script. Granted, that might make for a more complicated 
configure script...  But it might be interesting.

If it detects curses, on the system, it could built tn5250 (with 
cursesterm) if it detects X11, it could build x5250, etc.


> Finally, can anyone think of a way I could have the terminal start and stop 
> the trace at any time?  I'd like to add a button to x5250 that turns on the 
> trace and then turns it off, instead of starting the trace from the very 
> beginning of the program.

Isn't that just a matter of calling tn5250_log_open() and 
tn5250_log_close() at the appropriate times?

Though, I wonder if it's a good idea. The idea behind the trace file is to 
be able to re-produce errors. Being able to turn it off and on at 
different times might create a trace file that doesn't put the emulator 
into the same state as it would if they ran through the whole process.

Might be worth experimenting with, though