finishing write-byte-sequence patch

Sam Steingold writes:
 > that's because the anon cvs is run from the backup server with a 24
 > hour delay.  please do complain to the SF admins!
ok, have now sent a complaint about both that and
 cvs [diff aborted]: recv() from server cvs.sourceforge.net: EOF

 > 1. if you name your build directories "build-*", they will not be
 >    mentioned in the "cvs diff" messages with "?".
Thanks.  Maybe all this stuff should be collected into an
introductory guide for clisp developers.
(If you tell me now that this already exists I'll scream.)
Chapt 32 of clispnotes is the closest I see.
If I ever recover from this experience I'll try to submit something
to be added to that section.  Perhaps to be pointed to from faq.

 > 2. Please send the ChangeLog entry ASAP, and I will commit your
 >    patch.  Note that I will probably change some formatting, so your

I hope you'll also try to do something about the //'s and the ***'s.
At least eventually.

 >    "cvs up" will not necessarily be clean (especially considering
 >    Bruno's patch...)
You did once volunteer to supply this.
I suppose there are standards here of which I'm unaware, just like
everywhere else.

Here's a starting point which you may alter (or discard) as you see fit.

add :no-hang argument to ext:write-byte-sequence, also
gray:stream-write-byte-sequence 
Major changes to stream.d,
lesser changes to sequence.d, unixaux.d
minor changes to genclisph.d, gray.lisp, lispbibl.d, subr.d, subrkw.d,
unix.d

[Should the affected functions be listed?]

[Should I include remaining problems?]
- no-hang with buffered streams (buffered_flush can hang)
- input from buffered socket streams
- not clear that NO_BLOCK stuff is working as desired
 ...

I see also a need to patch impnotes.
I'll be happy to send suggestions for that later if you wish.

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf