Re: [OFF-TOPIC]

From: George Greer (greerga@circlemud.org)
Date: 06/17/99


On Wed, 16 Jun 1999, Daniel A. Koepke wrote:

>Everyone is focusing on getting 3.1 out the door before we worry too much
>about final decisions for 4.0.

True, but I like to collect things on
http://www.circlemud.org/~greerga/TODO.html in the meantime.

>And there are other options that, at least, George is investigating for
>4.0.

So far I've found glib (ftp://ftp.gtk.org/pub/glib/) to be designed fairly
well.  Ignoring the whole 'gpointer' and 'gint' nonsense of course.  I
don't see a problem with the 'gint32' type, but I really disagree with
'gpointer' as a type.  Reminds me of C++ references and how you forget when
you're working with a pointer and need '*' or not.  Otherwise, it finally
gives useful structured building blocks.  Re-writing a queue system for
every program does get tedious quickly.

>But, no plans are finalized and, as always, the final decision is
>Jeremy's.

Documentation is the current largest issue with the CircleMUD release.
There needs to be documentation written up and the current SGML was written
with a custom parser in mind that is no longer available.

>I don't think there's that much of a learning curve.  However, the ANSI
>standards committee working on C++ appears to be trying very hard to
>make the language harder to use and look at.  Point of reference: new
>style casting which use the template syntax, which was ugly and unreadable
>to begin with.

CircleMUD will always try to compile in a C++ compiler barring some stupid
incompatibilities in the language.  Their 'extern' treatment borders on
that.

>It should be done in whatever language and in whatever way that Jeremy and
>George determine best suits the future of CircleMUD.

JavaScript. :)

>I am pressing for less features to be in CircleMUD -- or to make the
>standard distribution features more easily removable.

I did just that in one of my code bases.  Boards, mail, shops, special
procedures, magic, skills, and a few other things were all compile-time
selected.  Any number of them could be removed independently restricted
only the obvious, like shops require special procedures.

I didn't do it just to be anal about things but was the end result of
throwing "#if 0" around liberally to isolate the code required by various
subsystems that was't with what required it.

>Not everyone wants the same implementation of races, classes,
>etc., in their MUD, and such creative minds should not be punished by
>having to rip out a mountain of code.

My method of board removal was basically "remove boards.o from object list
and #ifdef what breaks."  Wasn't too rough and most things were in the
right places when I did the other stuff.

>As far as bugs go, well...people don't typically release code with
>serious, obvious bugs in it.

bpl15 (and 16) currently have some real irkers.  Need to find the time to
make sweeping changes to fix them.

>Nor do Jeremy or George.  All of the 3.0s we've seen so far are betas
>that have been eliminating bugs.

"return (0)" notwithstanding. :)

That was a code consistency thing though.

>The next non-beta release, which is 3.1 for those keeping score at home,
>shouldn't have any bad, noticable bugs.

We could make 'version' crash, oh wait, that's been done...

>Of course, it's no trivial task to eliminate *all* the bugs...

void main(void)
{
  return 0;
}

Even the above has a bug.

--
George Greer            | Stock CircleMUD Bug Reporting or Help
greerga@circlemud.org   | http://bugs.circlemud.org/


     +------------------------------------------------------------+
     | Ensure that you have read the CircleMUD Mailing List FAQ:  |
     |  http://qsilver.queensu.ca/~fletchra/Circle/list-faq.html  |
     +------------------------------------------------------------+



This archive was generated by hypermail 2b30 : 12/15/00 PST