[Date Prev] [Date Next] [Thread Prev] [Thread Next] Date Index Thread Index Search archive:
Date:Fri, 16 Apr 2004 23:40:00 +0000 (UTC) 
Subject:Re: Open Source Poplog -- 'Official'? 
From:A . Sloman 
Volume-ID: 

Brent,

> Which reminds me:  What should be considered the
> 'official' state of Poplog?  I had high hopes that
> Jeff's sourceforge work would be the location for the
> current state-of-the-art Poplog sources, since this
> is a public location with CVS access that everyone
> can reference.

Eventually that should happen. Right now it is being used
mainly as the workspace for people interested in porting poplog to
windows.

Discussions of this have mostly been on the poplog-dev email list --
though that has been quiet lately: all the people concerned have to earn
their living doing other things -- so the work comes in bursts!

> I think it would be really useful to have a single
> master repository for the Poplog sources that could
> hold onto patches (like the one you forgot about) so
> they could be tracked by a release system.

Agreed. Jeff Best did add me to the list of people with access to the
openpoplog site, but I have so many different things to do and am always
so far behind with most of them them that I have not yet even learnt how
to use the system.

So when I have something to install I ask Jeff to do it, as he has
explained.

At present, there are so few people actually contributing anything to
the unix/linux poplog system (mainly because it has so few bugs!) that
it is simplest for me just to handle everything where I can quickly fit
things in without even switching from using Ved -- I use it for mail,
programming, documentation, reading news, running grep and managing its
output, comparing files, or directories, etc. etc. and sometimes there
is a problem that is holding someone up badly but is easy for me to fix,
so I just open another ved window, fix it, and then continue with what
I was doing, leaving a note for myself to update documentation
and the bugfixes directory later.  Usually this affects only add-on
libraries, e.g. rclib or simagent.

The work on windows poplog has two streams

    a. attempting to add graphics based on native windows graphics
       (being done mainly by Nico Aragon in Madrid, when he has time)

    b. attempting to get linux poplog to run under cygwin
        (being done/led mainly by Jeff, I believe.
        Non-trivial because everything has to be re-compiled
        from scratch using cygwin libraries -- in contrast with
        vmware, which alread supports linux poplog binaries under
        windows (at least in XP -- which is what I have actually
        seen -- but vmware costs quite a lot.)

   [I've only just noticed a web site for Colinux.
            http://www.colinux.org/
    I can't tell whether it provides the same functionality as
    vmware, e.g. running linux binaries directly, or whether it
    would require work to port poplog to it, as in cygwin..]

At present, if you are working on windows poplog, in stream (a) or
stream (b) it is best to use the openpoplog sources. Ask Jeff if you
wish to be added to the list of contributors.

Do you know anything about colinux?

Most changes I install are unlikely to affect the windows version of
poplog until it reaches the full functionality of linux poplog, in which
case I'll be able to provide updates from the files in

    http://www.cs.bham.ac.uk/research/poplog/src/new/

which are NOT well organised!

I don't install changes in the core poplog often enough to have worked
systematically.

The things I do change more often are the pop11 library-level add-ons
developed at Birmingham (rclib, rcmenu, poprulebase, simagent) and from
time to time David Young's popvision (which I fetch from Sussex where it
is used on Suns, rebuild for linux then install here:

    http://www.cs.bham.ac.uk/research/poplog/popvision/
    http://www.cs.bham.ac.uk/research/poplog/popvision.tar.gz

(it is due for an upgrade as David has made some changes recently.)

Sometimes Steve Leach upgrades objectclass. I then install that
in Linux and Solaris versions of poplog, and also make it visible
here for people who wish to browse or fetch and install in a running
poplog:

    http://www.cs.bham.ac.uk/research/poplog/src/new/objectclass/
    http://www.cs.bham.ac.uk/research/poplog/src/new/objectclass.tar.gz
        (Last updated March 2003)

I have also gradually modified the scripts and documentation
I use with the PC+linux poplog package assembled for our students

    http://www.cs.bham.ac.uk/research/poplog/bham-linux-poplog.tar.gz

making installation more automatic and allowing more options. As I am
not an experienced shell scripter it takes me more effort than it
should. I've also been gradually changing from csh to bash....
which I don't normally use as tcsh suits me fine.

> Also, I notice there are several cases where duplicate
> copies of scripts are present with certain features
> present or not, and certain paths set, etc.

Do you mean at the bham site, or the openpoplog site, or both?

Just to confuse things more, there is also the much older www.poplog.org
site set up by Graham Higgins and Steve Leach, which is supposed to
mirror the bham site, and add some extra pop11 facilities developed
mainly by Steve Leach, but every now and again the mirroring software
seems to get stuck and then people who use that site as a source for
poplog get into difficulties. I think Graham Higgins has been working on
it.

It also archives comp.lang.pop postings, but the archiving has recently
got stuck. Graham tells me there are problems caused by moving to a new
server which he plans to sort out.

> I'm
> curious why something like AUTOCONF isn't being used
> to allow you to have a few script templates that
> could be used to generate the final scripts actually
> used on a particular system?

The main reason that it is not being used for unix/linux poplog is that
I am the only person doing the packaging for linux/unix poplog and I
have never used Autoconf and may not have time to find out how in the
near future for the reasons given previously. I have never even
constructed a Makefile....

One problem I have found with a lot of stuff on linux tools is that it
is very hard to get good introductory tutorial documentation: developers
are often experienced users of things similar to the stuff they are
developing and they assume all potential users have a similar
background, which leads to over-terse documentation.

Another reason I have not done anything more general is that I have only
RedHat versions of linux to play with so I produce scripts that work for
that, then if people using other versions of linux complain, I simply
attempt to add either remedial documentation or extra modifications to
the scripts.

In the long run that's not good enough, I know.

If there were more bugs to fix with more people contributing, then
perhaps we'd have made more progress automating things.

I have yet to learn to use CVS, as I am often exhorted to do...

A very recent experimental change removed dependence on termcap, which
was causing problems for several non-redhat users. That was one of the
few things that required changes to core sources, a few weeks ago.

It seems to have caused no problems and made things easier for
Mandrake users, so I'll document that and install the patches
on the web site. They are already in the 'current' packaged
linux poplog versions.

[Incidentally, almost all the changes I make require help from someone:
E.g. Steve Isard made the crucial observation about how to remove
dependence on Termcap in linux poplog.]

Another parallel development is mini-linux pop11, developed by Steve
Isard, and installed here
    http://www.cstr.ed.ac.uk/~stepheni/poplog
or
    ftp://ftp.cogsci.ed.ac.uk/pub/stepheni/poplog

and copied here:
    http://www.cs.bham.ac.uk/research/poplog/mini-linux-pop/

Described thus in the AREADME file:

    This version is particularly intended for small Linux distributions
    capable of running on systems with older Intel processors and
    limited memory and disk space.  Because such systems are frequently
    based on libc5, which is smaller than the now current libc6, the
    executables here have been recompiled with libc5 libraries.  They
    will also run on later libc6 based systems that include libc5
    libraries for backwards compatibility. I have run the pop11 here
    under RedHat 6.0, 6.2, 7.0 and 7.1, as well as on the libc5 based
    distributions
    Mulinux (http://sunsite.dk/mulinux),
    Monkey Linux (http://www.ibiblio.org/pub/Linux/distributions/monkey)
    and
    Tomsrtbt (http://www.toms.net/~toehser/rb/).

He provides a tiny reduced pop11 package that fits on two floppies!

I have no idea how many people are using that. I guess if anyone finds
problems they tell Steve and he fixes them. But it has not changed
much.

Steve copies core bugfixes from Birmingham - e.g. a bug in array bounds
checking, and a bug in compilation of complex conditionals.

I guess that like me he finds it most convenient simply to work on his
own machine. But eventually this should also be added to openpoplog.

My feeling is that the time for that will come when there's a useable
full poplog for windows in the openpoplog site. Then we can make that
the master site for everything.

Does that make sense?

Aaron
Ps
I have made some unsuccessful attempts to get research council funds for
someone to help me with poplog support and development. If I had
such funds and had someone working full time, even if it were only for
a couple of years, then things could change a lot.