Compo returns

have a look at this

Bloody hell, that looks nice.

And as a person who has a sideline in fighting brambles, I don’t want to even think about how much upkeep would be required.

I have a copy of the printed manual here. It’s approx 50 double-sided A5 pages (100 sides).

Page 1 says “Copyright © APDL and Rob Davison 2005”.

I’d be happy to scan it to a PDF. Just say the word…

Compo and Ovation Pro “Install the OPCA applet…” – if I was scanning a lot of books, maybe set up a camera.

Erm, “The Word”. :-)

My copy has a date for “1994,1995” rather than “2005” so yours may be more up-to-date in content.

FWIW I have tried recompiling the "lightsquare’ ‘C’ prog that gets used and this falls over on my current machine with a “division by zero” error. The old version doesn’t work on my new machine, but works fine on my old RPCEmu setup. If anyone can diagnose that, please say. Afraid I’m a lousy programmer and also have forgotten much of what I knew when these scripts were written!

Here is a demo scan of cover, title page, Contents, Introduction, plus page 7 which has a colour illustration.

Alas, with !Netsurf that gives me a blank page of HTML!

Looked at the code on the page and can see a munged URL but not found way to decypher the ‘true name’ and fetch it.

BTW On the only other webforum I use they have a ‘personal message’ system where forum members can talk off-forum. Does this forum have a similar capability?

The link to the Compo scans looks OK in Iris 1.029, its a link to a PDF download on the pCloud server. It does not make available the PDF download in NetSurf here either.

My apologies. It’s stored on pCloud whose secret links are not compatible with Netsurf. However, I have now put a copy in a pCloud public folder here which Netsurf can download.

Thanks. I’ve now got a copy of what you scanned.

I’d certainly welcome a copy of the latter parts on scripting. My manual is for an earlier version which didn’t cover that. However there is a possible need to delete the “APDL” from the cover, etc, in any copies made public.

So I’d welcome a copy being available with that caveat.

Also: I’ve had a GOOD afternoon! :-))

Rob sent me a copy of at least some source material and various versions, etc, back c 2000. He has lost his copies. And although I have in the past repeatedly searched my old backup DVDs, etc, I couldn’t find what he sent to me. HOWEVER I JUST FOUND a set of the old items he sent. Real joy to find them and that the DVD hadn’t ‘faded’ or got lost.

Lots of stuff. I’ve sent a copy to Andrew so he now also has one. Not sure how much it covers, but with luck a helpful find.

This is weird. I just send the above material I’d found to Andrew and he can’t extract zips from the zips! Yet they are fine here when tested. Too late to do more as he’s off to the show.

Sounds like Drag and Drop Archive – phone Pdf reader doesn’t like
reading PDF file inside a zip – got to get it out first :-(

I’d certainly welcome a copy of the latter parts on scripting.

Yes, as you’ve spotted, scripting is described in the 8 pages of Appendix 4. Along with, of course, the HTML of Composcript 1.22f on IconBar and on the Compo CD-ROM.

a possible need to delete the “APDL” from the cover, etc

Okay, that can be done.

After Appendix 5 Keyboard shortcuts is a further section of 22 pages on PCA and PCA Applets (APDL) which should be included.

My CD-ROM contains Compo 1.23c. Just a shame that it doesn’t contain the files from which the manual was printed. Are they not among the files you have from Rob?

No. I don’t think Rob had the files for the manual.

Update:

I have now copied the files I have onto an ADFS SD card and I’ll post it to Andrew when I can – hopefully later today or tomorrow. Not ideal, but with luck it will get to him and RM won’t lose or smash it!

As you’ll know the main code for !Compo is ‘squashed’ BASIC which makes it a PITA to understand. But Rob did also give me one (current at the time) unsquashed BASIC version as well. This will all go to Andrew on the card. I’m hoping this will help anyone trying to update or help with the project.

My main puzzle at the moment wrt this is why some scripts I have will work on my RPCemu but not with Aemulor on my ARMX6. Maybe someone else can diagnose that and explain… Particularly annoying for one I wanted to use for graph plotting by poking to a sprite. Method works on RPCEmu, but not on my ARMX6 – regadless of deploying Aemulor.

I have send a quick fix in the squashed Basic RunImage to Andrew. So it works with more than 2GB RAM machines.

I’d be interested in how people get on with this script.

http://jcgl.orpheusweb.co.uk/Compo/plotblank.zip

It (should) load a gif and then use it as a place to plot some patterns.

It works for me on RPCemu, albeit not always, but not work on my ARMX6. Results also seem unpredictable. Note that if it does work you may see an hourglass for a while. My puzzle is being able to get consistent success and know what setups it can’t work on, or can.

I’m running ROD 5.29 OS on my ARMX6

The return in interest in Compo and reviving+updating+expanding it has also prompted me to look back at my old email files. Using !Pluto I’ve kept copies back to the mid 1990s. However the older archives are rlists whereas the later ones are ‘external boxes’. So I’ll need to use the rlists to produce some ‘older’ external boxes which make searching with !Pluto far easer and more powerful.

I’m now tempted to do this as even a short search recovers some forgotten conversations with Rob D. about the coding and also about possible reasons !Compo became unobtainium. Reasons that are now irrelevant so wouldn’t be a drawback.

That said, if I did recover the ‘full story’ I’d need to ask Rob to ensure he’d happy with what I then made public about it.

Another thing I’ve found is a ‘reference’ HTML set of info on composcript which lists and explains the commands. That may be useful to put up on the web for people to see and use. Or is that already somewhere public? It being on the web would bypass the current lack of a public manual about it.

@BernardBoase Nice quality of the scan! If you complete it, might be nice when you remove also the visible paper holes.

When I rescued Ray Favre’s “Dr Wimp’s Surgery”, the format of the original book was such as to have a wide inner-margin on each page to accommodate the spiral binding.

Because I aimed to create (as closely as possible) a facsimile of the original, I produced a version of the Impression file which showed where all 16 of the holes for the binding were on each page – a trivial job done just once in the document’s master page.

In the end, because the intention was to re-publish it as a PDF, I didn’t proceed with the ‘holy’ version, where physical holes in an electronic document aren’t really appropriate…

…unless someone wants to produce it on punched tape.

I might be creating a hostage to fortune but, if you can get the original scans to me, Bernard, there’s a strong possibility that I could OCR the document. Having the text available for editing would give an element of future-proofing to the Compo project.

You might still have my email address so let me know what you think.

I’ve now emailed the more ‘significant’ files Rob gave me to Andrew. That includes a ‘non squashed’(i.e. source) version of the main programme in BASIC, plus a fleet of test/demo scripts and the files they manipulate.

The full set (34 Megs worth) I’ve put onto a CDRom that I can post to him if we can’t find a faster way via something like FTP. ( I am avoiding zipping any of this to dodge another ‘zipped zip’ problem.)

Testing these script files is difficult for various reasons. Many can’t work on my ARMX6, but may on my RPCemu. Some work ‘sometimes’ because I’ve set up something that wasn’t set up previously, etc, etc). Also may depend on the version. (The files include a number of versions of Compo, but all with compressed basic main progs. The one uncompressed version is separate.)

One that surprised me is a ‘presentation’ script. This lets Compo act as ‘presentation software’. It goes full-screen and uses objects on the canvas as ‘forward’ and ‘back’ buttons to step through the presentation. And being compo can also trigger animations, etc. I was really astonished by this. Given an extension it could probably play audio on command as well!

Snag was that on my ARMX6 it dropped the screen mode to being so small that the right-hand icons on my icon bar were off screen when the compo presentation was quit! I’d totally forgotten compo could do this sort of job, but it became obvious when I tried it!!

In effect, the scripting will also run user-written code. e.g. the ones where the process called was written in ‘C’ to number-bash some data and then show the results. The composcript calls the compiled result and shows the visuals created.

‘zipped zip’ problem

If you need any help with zips let me know. Presumably you have a choice of SparkFS (which is re-entrant and can handle zips in zips) or InfoZip which should have no problem at all. Or is it what has happened to the zips on route.

I used SparkFS 1.46 (06-Jul-20) on my ARMX6. Would a later version be better in this respect? I then attached the zips to a set of emails to Andrew. I use !Pluto 3.20 (20-Jan-24).

The files I zipped contain some old zips. Here they seem OK. But Andrew said they gave him problems.

Don’t have InfoZip, so more on that would be welcome!

FWIW Just been experimenting with some of the old composcript programs on RPCemu. They mostly work nicely once I set things up so Compo can find ‘helper’ progs like those for gif/sprite conversions, etc. But I suspect I need to shift that testing from ‘HostFS’ to ‘HD4’ as some things aren’t right. Not used RPCemu seriously for some time so have to relean a few things about that as well!

If you could send me an exact zip Andrew had problems with…

A thing you can do, is create a zip, add stuff, and then rapidly do something with it – email it. Result is that the zip has not been closed and is corrupt (missing central directory). Your copy on disc sits there and gets closed in due course and is fine.

Person at the other end gets a corrupt zip, even though your copy is fine. Their solution is to use the SparkFS zip module command “ZipUseCentralDirectory 0”, (check that) which makes the module use the headers rather than the central directory.

You can change the timeout SparkFS uses – make the time shorter.

That is speculation. Perhaps something else has happened.

!InfoZip seems to come as standard with RISC OS these days,copy in utilities or somewhere else. If not this is the official link:

http://www.starfighter.acornarcade.com/mysite/utilities.htm#infozip

It wasn’t so much that the zips themselves wouldn’t open, but upon attempting to load the source files (zips inside zip etc) then any text editor I tried would crash. SparkFS would then report memory corruption. I was in a blind rush/panic pre-show though, so didn’t explore too deeply beyond “well, that’s unpleasant” and quit/reloaded sparkfs and did other things.

I’d probably drag the inner zips out to disc then try their contents. Which is what you’d have to do with InfoZip.