I've managed to produce a script available here

        https://www.cs.bham.ac.uk/research/projects/poplog/V16/getpoplog.sh

viewable in a browser as:

        https://www.cs.bham.ac.uk/research/projects/poplog/V16/getpoplog

On my fedora machine at home, with a fairly fast internet connection, if I
run the script it fetches all the required files and builds a usable poplog
with the documentation files and packages from Birmingham included -- in
less than a minute. I was amazed at the speed.

I have found out how to get an up to date version of Waldek's poplog from
github (as a zip file). I had to fix a couple of minor things (e.g. setup
scripts for tcsh users, and re-packaged everything as a tar.bz2 file,
following Waldek's example, but using a different name:
        
http://www.cs.bham.ac.uk/research/projects/poplog/V16/latest_poplog_base.tar.bz2

The getpoplog.sh file now fetches that and also some tar files with
additional poplog packages and documentation.

It then untars the latest_poplog file, creating poplog_base, and runs the
script in that directory that Waldek provided, except that I had to convert
it from bash to a tcsh script to make it run when invoked from the download
script.

I found no way to get the download script to run the bash script: I had to
manually 'source' it instead. I never worked out why.
(Perhaps a security feature in bash?)

Anyhow the tcsh version worked and rebuilt the basic poplog, so I extended
it with commands also to unpack the extra documentation and packages files.

So now everything is fetched and installed using one command -- and on my
machine the whole process, including all Waldek's compilation to rebuild
poplog, takes at most about a minute. I have not timed it!

[My shell programming is very primitive. It's possible that someone could
produce a better fetch+install script. Mine is full of comments to help
if anything goes wrong.]

SUMMARY:
Everything is now fetched and built by downloading and running the
getpoplog.sh file.

After that the poplog_base directory has a file called USEPOP which has the
contents to be set to $usepop. E.g. using tcsh and bash respectively:

        setenv usepop `cat poplog_base/USEPOP`
        export usepop=`cat poplog_base/USEPOP`

After that, bash users do the following to set up search lists:

        source $usepop/pop/com/poplog.sh
        
tcsh users do:

        source $usepop/pop/com/poplog.csh

This is all included in
        
        
http://www.cs.bham.ac.uk/research/projects/poplog/V16/latest_poplog_base.tar.bz2

Documentation of the process, with other background information is in
        
        https://www.cs.bham.ac.uk/research/projects/poplog/V16/AREADME.html

The linux prerequisites for redhat-based and debian-based linux are linked
from the AREADME file

        
https://www.cs.bham.ac.uk/research/projects/poplog/V16/AREADME.html#prerequisites
        
I'll add any others if information is provided.

I presume that eventually all of this can be handled by github.
(I wish it preserved dates of unchanged files .... to help with debugging.)

Testing espeak version of Eliza:

If you have installed the linux espeak library, then after the above
installation is complete you can do this:

        mkeliza-speak

ignore the print out
        Please type
           eliza();
        [the library needs to be altered slightly.]

Turn on your speakers, then do
        pop11 +eliza

Enjoy your conversation. Kids love it ... for a while.
Type in anything -- but no line breaks. Questions, statements, commands
will all produce responses (possibly incoherent). Eliza's rules are easily
replaced/extended. There's no learning, except that occasionally the user's
input is stored, and re-used later.

A future project for someone: replace keyboard input with spoken input.

-----

To test the winograd-inspired simulated robot -- create the saved image:

        mkgblocks
Ignore the stuff printed out.

Then run it:

        pop11 +gblocks

it asks if you want to use xved, answer 'y' or 'n'.

If necessary remove
        $usepop/pop/lib/psv/gblocks.psv

or other unwanted saved images. They all depend on 'startup.psv'
which cannot be removed, though if needed it is easily rebuilt,
using:

        mkstartup

And now I have to try to catch up with a huge backlog of other things.

----------

Waldek's (32-bit) port to ARM is available  here:

        http://www.math.uni.wroc.pl/~hebisch/poplog/poplog_arm.tar.bz2

If anyone tries it, please report here.

Thanks.

I wish happy festivities to all!

Aaron

Reply via email to