On Mon, Sep 21, 1998 at 02:26:25PM +0200, Hrvoje Niksic wrote:
* Why is the syntax of declaring dynamic SUBR's different from
the
rest of XEmacs? I'm not sure this is acceptable, really.
I fixed that. There
was a reason at the time, I understand a few things better
now and its gone away with the new patch (the one to symbols.c).
* Why are you using specialized error functions instead of the Emacs
ones? As a reminder, Emacs has error(), signal_simple_error(), and
a bunch of others. See eval.c for details.
For a good reason ( think). When I
need to signal an error I need to make
sure that I unwind any modules that were partially loaded or that were
loaded as part of a dependancy chain, and I need to do that before I
call the error functions, which will asynchronously terminate my code,
leaving things in a mess. I could simply call error() but I wanted the
control. I need the varargs stuff to get the error string, and I COULD
simply build up a string and call error(). Thats no different to calling
signal_error() though, so I dont really see the problem.
Also, your error messages are at odds with the usual Emacs error
conventions. Please look them up, or mail me privately if you need
more details on what exactly I mean.
I need more details.
* I believe your code is not Mule-clean, and as such not eligible
for
application to XEmacs. Hint: casting XSTRING_DATA (foo) to `char *'
is something that should *NOT* be done. Again, ask if you need more
explanations.
Yes, please. I am not sure what the requirements are but things can
be
changed easily.
* dll-list-modules should really return a list, not a vector.
Vectors
should be used only for data structures that require random access,
as opposed to front-to-end searches.
Done.
Everythings in the new attached tar file. Applying this to a virgin b55
tree should work seamlessly. Please check this out and let me know what
you think?
Kean.