Re: mudlog <01bcdb3a$db8bf560$>

From: Jeremy Elson (jelson@CIRCLEMUD.ORG)
Date: 10/17/97

"Kenneth G. Cavness" writes:
>Jeremy, if you do any more new functionality before you make
>3.0 go out of beta, make file-logging runtime please?

Definitely.  In general, I don't want to add any new functionality
until 3.1, but this is serious enough that I'd consider it to be a bug
(Windows users can't redirect their logs like UNIX users can, so they
just lose logs.)

Yes, it's been like 18 months since I've released code :-(, but for
the last few weeks I've desperately been trying to finish pl12 in
between writing a 75-page user manual for software I wrote at work and
studying for the CS subject GRE's that I'm taking in 2 weeks.

pl12 (and all future pl's until non-beta-3.0, which will be called 3.1
to avoid confusion) will solely fix what I consider to be bugs, which
includes both "real" bugs and features whose absence is serious enough
to be considered a bug (e.g. aliases not saving).  Also I'll be
incorporating a lot of bug-patch code that people have been sending
for the past year and a half, all of which has been neatly filed away
in a directory for the time at which I'll finally be able to do
something with it.

Also, I've enlisted the help of Mr. George Greer to get 3.0 done.  He
has written a huge number of patches, both bug-fixes and feature
enhancements, and I've asked him to personally take part in modifying
Circle's code by adding some of his patches to it.  Again, for now,
this will only include bug-fix patches, but feature patches may be
included later.

Also, I'd like to reiterate my design goal of releasing a totally
rock-solid, bug-free, WELL-DOCUMENTED MUD base.  I have no intention
of ever adding races, for example; as far as I'm concerned, that's
still an add-on feature.  On the other hand, I expend a great deal of
effort on doing things like making sure Circle prodces the minimum
possible network traffic -- this is something very fundamental that
all MUDs should have, and that almost no MUD imp will ever work on.
I've just finished code for Patchlevel 12 that will reduce the number
of TCP packets generated by the MUD (and generated by clients) by
30-50% -- I consider this to be the perfect way to spend my time,
Circle-coding wise.

(BTW, this was done by rewriting the prompt-generation code, so that
prompts are written to the network as part of the same system call
that generates the output of the command that generated the prompt.
On most OS's using write() twice generates two TCP packets, and
produces two ACK-bearing packets from the client.  Now only one
packet is sent and received.)

Also I've rewritten some of the output buffering and overflow handling
so that it's never, *ever* possible to be dropped due to an output
overflow.  (Type "where sword" all you want -- you may lose a lot of
data and see "** OVERFLOW **", but you'll never lose link.)

After finishing the docs, integrating some more bug-fix patches, and
adding some of these critical features like alias-saving and whatnot,
3.1 will be released in all of its 5-year-late glory :-).


     | Ensure that you have read the CircleMUD Mailing List FAQ:  |
     | |

This archive was generated by hypermail 2b30 : 12/08/00 PST