dclemans@mntgfx.mentor.com (Dave Clemans) (11/04/87)
In case there's interest, I've just managed to get the recent release
of "C" news succesfully installed on an Apollo system. Relevant
details: ring with a couple hundred Apollo's, Aegis 9.6 & 9.7,
Domain/IX 9.5.
Performance is excellent. (Guessing) I think it's about 10 times as
fast as the version of B news 2.11 we were running previously.
The only real problems were:
no overall installation script; it took a while to put everything
into place
there were some garbage files in our news spool directory that
caused "histinfo" to crash while building the new history file
there is a conflict in the "input" subsystem between what it
expect relaynews to do and what it actually does. I solved it
forcing the "-p" flag on by default in relaynews so that it
always writes to log files instead of to stdout/stderr (writing
to stdout/stderr caused the newsrun shell script to think that
relaynews always failed)
there is a conflict with the "putenv" subroutine; Apollo provides
a "putenv" that matches Sys5 semantics (specifically it returns
zero on success) (and for Apollo's you need to use that putenv
routine). But the putenv that comes with "C" news has the
opposite return semantics (non-zero on success), and since that
was what was wanted by the error tests at the beginning of relaynews,
relaynews was exiting immediately with no messages. I fixed
this by reversing the sense of the error tests.
dgc