RE: Use Tcp.h?

[Harold L Hunt <huntharo at msu dot edu>]

Nicholas,

> Anyhow, I think we are missing the whole point of this thread, what were
> *YOUR* findings. 

I forgot to draw attention to what I found, but I did post your build snippet
with warnings and my build snippet that didn't have warnings for the same file.

My overall results were that I got no new errors or warnings.  I did a 'make
install' and ran the server with local clients as well as with a -query to a
KDE machine with no problems.

> Again, I'm sorry for not sticking to it exactly.  So what were your
> findings from your build?  What is your conclusion?

The conclusion is that we might as well remove -DNO_TCP_H and the #if
!defined(__CYGWIN__) from whatever file I said it was in.  (Note that
-DNO_TCP_H was present in the initial version of cygwin.cf... so it is just a
define that has not been noticed as uneeded until now.)

> > On a side note, I find it hard enough to remember all the
> > builds/flags/compilers/etc that I'm using.  I don't have any space in my
> > brain to store state information for other developers.  You have to feed
> > me some details everytime you ask a question, else you can assume that
> I've
> > forgotten those details.
> 
> 
> OK, next time I'll be better.

Appreciated.

> However, are you aware that the CVS HEAD isn't even building period with
> regards to your cross directions?  It fails in the initial stages, because
> it seems that some of the macros you have defined in your host.def are no
> longer valid.  If you do not believe me, try it yourself.  It seems that
> they have mucked around with the cross building configuration files and
> rules files.  We better get on them about that, too.  I believe that some
> stupid SuSE developer is responsible for the whole mess...

Oh, I believe you.  But like I said in another post (which hadn't been written
when you wrote this), I'll have to do a substantial amount of work to my Linux
machine to be able to do a cross compile of Cygwin/XFree86.

> P.S. - When you get a chance, you should read those other posts I made,
> especially regarding the conflicting files in lesstif and XFree86-prog...

I read them, but you're going to have to do more than just suggest what to do
with the host.def files.  I hope you realize that your simple, ``should we
remove -DNO_TCP_H'', question has cost me about 5 hours already in looking at
source files, doing build tests and writing detailed correspondence to the
mailing list.

Now you are asking about Lesstif's host.def and our host.def and all I can see
is that our host.def is empty so I can't see what problems it will cause for
us not to do anything with the host.def files.  Furthermore, I don't see how
you could even fix this with pre-remove and postinstall scripts.  I mean, how
are you going to determine which host.def file is installed, how are you going
to determine if you need to remove the currently installed host.def file
(maybe package Z's host.def file was already overwritten by another package
installation), and how are you going to determine which host.def to install in
place of the one that are you are removing?  Not to mention what sort of
naming/storage convention are you going to use to identify the original
host.def files that come with each package?

So yeah, I read your post and I saw that it raised more questions than it
answered, so I forgot about it.  I'm leaving this one up to you, or somebody
else, to figure out.

You have to remember, as I've said time and time again, I'm a horrible X user
and I'm even a horrible X developer.  You see, I don't have years of
experience with hundreds of X programs and with hundreds of X libraries.  I
only have experience with X Server implementation and in that I only have
experience with X Server development for Cygwin.  I just don't have enough
experience to solve questions about host.def files easily.


Harold