Re: Would you accept a small patch to make window(1) work on non-BSD systems?

[Jason Spiro <jasonspiro4%gmail.com@localhost>]

Jeremy C. Reed <reed <at> reedmedia.net> writes:
> NetBSD's source does include various portability code for non-NetBSD.
> 
> Send your patch.

My work isn't ready for production yet.  The rest of this message explains why 
not.

My work so far has been a quick hack.  The Makefile does not make 
intermediate .o files, just a single binary.  The two code changes I made were:

-  I commented out "#include <util.h>" from wwpty.c.  But that's needed on 
NetBSD for the openpty function prototype.

-  window(1) uses the historical BSD signal API.  I made a typedef defining 
sig_t in wwsuspend.c, since BSD defines sig_t in the system headers but most 
other Unices don't.  I don't know what wwsuspend.c does, and the README's 
internals primer doesn't explain it.  But I bet that because of what I did, 
wwsuspend.c probably won't work on other Unices.  The ideal would be to switch 
window(1) to using POSIX signal handling but I don't plan to do that.

I plan to wait for a reply to 
http://thread.gmane.org/gmane.os.netbsd.devel.userlevel/9839/focus=9856 . Then 
I'll improve my GNUmakefile further and use ifdef's to implement the 2 code 
changes above for non-BSD systems only.  I suspect that using autoconf plus 
automake for feature detection and header dependency tracking would be 
overkill here.  But I can't assume users will have mkdep.  So I'll use the 
dependency tracking method proposed in 
http://make.paulandlesley.org/autodep.html#output .  If you have anything to 
add, your comments would be appreciated.

Cheers,
-j
jasonspiro4 at gmail.com