>>>>> "Rob" == Robert Nation 885-9815 <rnation_at_mansvr.sanders.lockheed.com> writes:
Rob> Some of you may have noticed that it is not hard to add functions
Rob> to fvwm. TogglePage isn't the ideal example, because it tinkers
Rob> with a few functions outside of its own execution, but I'll go on
Rob> anyway.....
Rob> Most unix systems these days have the option to dynamically load
Rob> object files to a running program (gnu:dld, sun: dlopen/sym/etc,
Rob> OSF1:dlopen/sym/etc) has anyone thought of making most of the
Rob> functions dynamically loaded modules? They could be loaded on
Rob> demand - if you don't find a function, you can dynamically load
Rob> it if you can locate its .o file.
I've thought about this from time to time, since I implemented dynamic
loading (under AIX, SunOS, Solaris, and HP-UX) in the program that I
work on for my job, and it's been extremely handy in our product.
But I thought that the module implementation was probably more than
good enough for fvwm (although perhaps some additions to the protocol
might be useful) and more portable, so I didn't persue it much past
thinking "hey, that might be cool".
Rob> I thought of doing this a long time ago, but it adds complexity
Rob> to the build-process - since you need to accomodate systems that
Rob> lack the dl* functions.
Making it portable might not be too bad. It could be done with
generic interfaces to the dynamic load functions, files with specific
implementations, and some defines and stub functions for systems that
hadn't been ported to yet. It would probably be easier doing it via
AutoConf instead of Imake as well.
Well, I'll add it to the TO-DO list to investigate at some point. I'd
be able to cover Linux, AIX, SunOS, Solaris, and HP-UX.
Chuck
--
To unsubscribe from the list, send "unsubscribe fvwm" in the body of a
message to majordomo_at_hpc.uh.edu.
To report problems, send mail to fvwm-owner_at_hpc.uh.edu.
Received on Mon Jan 29 1996 - 10:29:28 GMT