[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

CVS ??!!??

From: Keith Hopper
Subject: CVS ??!!??
Date: Mon, 26 Mar 2001 09:03:47 +1200
User-agent: Pluto/2.02b (RISC-OS/3.60) POPstar/2.02

     In answer to Nobbi's amazement about my not using make - if someone
else has written it and nothing goes wrong than I am OK - otherwise I just
do fixes by hand - so much quicker than baffling oneself about some of the
most obscure sequences of characters that I have ever seen!

     I have no objection to a CVS approach - provided someone can point me
to a BOOK which might help me understand it.  I already see terms like data
base and diff in the messages.  The only thing I can ever do when someone
send me a diff file is to eyeball it with the two offending files to work
out what changes have to be made -and- far more important, WHY they have to
be made - and make them - much faster than using a bit of software you
don't trust and hoping like 'h**' that the other person knew what was going
on and had done the correct thing in the light of the design philosophy.

     Even Nobbi occasionally puts in a fix which I have to immediately
remove - 'cos, Nobbi, you look at files you have already written and want
to 'fix' things to work as before.  For example, you keep on throwing in a
'has' routine which I have carefully removed/changed to contains everywhere
- if only to make things consistent across the whole library - but more
important just on the grounds that no container has things - it does,
however, contain them - and, of course the same is true for sets as maps as
lists as ...

I have updated the Maths stuff in my master copy - which all seems to work
- as I said earlier I understand number systems and numeric models and so
on but maths libraries give me a frisson which makes my toes curl up (so to
speak!).  I have also incorporated some stuff which I got from someone in
Japan - to whom I have been trying to send my thanks for days - only to
have the mail system hang up on me - this morning I -think- it went - but I
am only keeping my fingers crossed.

     May I strongly make the suggestion/request that someone really sorts
out the concurrency stuff - which seems to have problems with the garbage
collector conflicting in some way with OS libraries (eg pthreads).  The
general philosophy is to produce one light weight process per CPU and build
active threads on top of that.  It sometime works and sometimes doesn't -
so I would like some other head to see what stupid mistakes I have made!!!!

     Only when that is done and the full library tests happily go through
the loop (as in aerobatics!) would I be happy to see further open
development.  At least we then have a sound, well-tested, basis (apart from
the inevitable 'last few' bugs) on which to build.

     Does this sound like a suitable initial starting point for
co-operative development??

     As a sort of aside, I have a student trying to build a windows and
graphics library for speed and portability using (initially) the X window
manager interface (raw X).  The graphics is designed to use OpenGL - the
entire interface is being worked on in the hope that it will support SVG
more or less directly with suitable parser/generator add-ons.  Don't hold
your breath, but the end of this year seems a likely first version target
date.  Once that is built then OS specific window manager stuff will be
developed in order to provide the lowest possible overhead.

     Some of you may have spotted the RTE directory in the latest
distribution - it contains more or less complete alpha test code for a bare
bones run-time engine - which needs further integration testing.  Once we
have hammered it to death then our intention is that a Sather thin
interface is built of calls to be generated by the compiler.  It
incorporates at the bottom level a multi-threading basis for
interrupt/exception handling - in an absolutely uniform way - a collection
of garbage collector threads sits on this as does a distributed interface
message system - the whole at run-time about 20k as far as we can estimate.
At present we don't have a warm body to finish it off (he left for pastures
greener - well with greenbacks as our US friends might say) although this
might change later this year.

     Apart from that I am personally continuing work on the specifications
- well on the first version (no doubt with lots of bugs) so that when
effort is available to build a new compiler we have a definition to work
from.  I am also writing a tutorial 'book' in HTML for use in teaching here
- but it might also be of interest to a wider audience.  Just let me run it
past a couple of courses before the end of the year and get theitr comments
built-in and we will see.

     Well!  All I can say now is 'over to you'!



City Desk
Waikato University
[PGP key available if desired]

reply via email to

[Prev in Thread] Current Thread [Next in Thread]