ROOL Forum: Recent posts

Fri, 31 Oct 2025 22:38:24 +0000

And a new procedure call standard and library to support that and…

softfp would be compatible, wouldn’t it? I mean, it won’t be the most performant, but it’d link with anything compiled for the fpe as long as you had vfp hafdware (and reversed the words)?

Fri, 31 Oct 2025 19:16:26 +0000

It also says it’s RogueWave C++ libraries – where’d you see the apache reference?]

https://lists.apache.org/thread/yl8p36x997bwq3510byl91…

Fri, 31 Oct 2025 18:39:00 +0000

Adding vfp support looks pretty simple, though I probably won’t bother as it’d need lots of testing

And a new procedure call standard and library to support that and…

Fri, 31 Oct 2025 16:30:57 +0000

Thanks Steve! Clearly anticipating further releases ;-)

Fri, 31 Oct 2025 16:28:22 +0000

You can now download this from https://croftnuisk.co.uk/coltsoft/ or wait a little until it reaches PackMan’s updated list…

Thanks for continuing to develop PipeDream (and Fireworkz)!

It looks as if the download link on the website for 4.63.00 should end _00 and not _02, though…

Fri, 31 Oct 2025 15:28:46 +0000

You can now download this from https://croftnuisk.co.uk/coltsoft/ or wait a little until it reaches PackMan’s updated list…

Source available in GitHub for the curious – https://github.com/skswales/colton-software-pipedream4 – as well as downloadable binaries in case my site goes phut.

Fri, 31 Oct 2025 14:38:05 +0000

It looks like the compiler was further developed up to ADS 1.2

Ah, it wasn’t clear to me if ADS was the same as SDT, though I didn’t investigate. That change log implies they may be, but I’d need to read something that said so more explicitly. I recall ARM bought Keil, a compiler vendor, and replaced Norcroft with that, and then they later replaced Keil with clang.

The SDT C++ library is from Rogue Wave. I don’t think it’s been open sourced, but you can find copies (at least, of the headers) on the web. I don’t think it’s worthwhile investigating as it’s pre-namespaces so not useful. I wrote a replacement for CFront’s iostream (albeit with a massively cut-down API for Bill’s library) in about five minutes, just calling stdio, but I hate iostream.

I was told that the ncc on the ROOL site was effectively a dump from Lee Smith’s laptop when he retired (which explains why there are quite a few duplicated files with different names, and two different files to handle C++ method overloading). I was also told it’s probably SDT 2.11a, not necessarily 2.11a (as the readme says).

Adding vfp support looks pretty simple, though I probably won’t bother as it’d need lots of testing. It’s just a case of mapping the JOPCODE for double add to a the double add instruction, etc. I don’t think the inline assembler supports any fp, so I’d not bother with that. I’m sure there’d be problems along the way…

[EDIT: I hate ARM’s documentation website. But it does, indeed, say ADS is armcc/armcpp. It also says it’s RogueWave C++ libraries – where’d you see the apache reference?]

Fri, 31 Oct 2025 11:53:14 +0000

It’s certainly not more up-to-date than ROOL’s as it stopped being developed 25 years ago. For the same reason it doesn’t support VFP (though it does support the AMP, which is an obsolete FPU developed by Intel).

It looks like the compiler was further developed up to ADS 1.2. (vfp and armv5te but elf only)

https://developer.arm.com/documentation/dui0064/d/diff…

I also found this, https://github.com/trapexit/3do-devkit/tree/master/bin… which is the compiler from SDT 2.51.

The C++ library that it uses was open sourced but looks like the version is too recent. https://stdcxx.apache.org/download.html.

Fri, 31 Oct 2025 08:11:01 +0000

For example, is it more up to date with how C has developed?

It’s certainly not more up-to-date than ROOL’s as it stopped being developed 25 years ago. For the same reason it doesn’t support VFP (though it does support the AMP, which is an obsolete FPU developed by Intel).

C mode has ‘__inline’ (which a system header could define to ‘inline’), and 64bit int types, but curiously ‘bool’ is guarded by a C++ ifdef. Given ‘inline’ is too, as an alias for ‘__inline’, it’s probably worth trying the same with ‘_Bool’.

I see no reason why it wouldn’t offer all the features of the SCL, but I’ve not tried them. You’d probably need -apcs… /fpe3, too. I suppose its main advantage, for some, is that it can cross compile from any other OS.

I’ll release my patches once I get a chance to tidy up my changes, which are a mess as I was learning how everything fits together.

Fri, 31 Oct 2025 07:48:02 +0000

I don’t care that it costs a trivial amount of money, I care that it is shit debugging experience

I came up with a plan to make it properly multitasking using wimp filters. I convinced myself it would’ve probably worked, but I never got round to it.

Trouble is, no one at Acorn really used DDT, so it never improved. Stuff was all printf-style debugged with TML podule cards (I gave mine to the CCH: https://www.computinghistory.org.uk/det/33949/TML-Podule/ ), and they didn’t really write many apps until ANC started porting things like Java, Macromedia, Citrix… (which is why we ‘upstairs’ had compiler conversations as the world was moving to C++), and updating Fresco and then Browse.

But, curiously, before I joined Acorn I, erm, ‘stumbled across’ a massively updated, multi-source-window version of DDT, claiming © Acorn – it appeared to be an upgrade rather than a rewrite. It didn’t exactly solve the ‘shit debugging experience’ problem, but it was a big improvement. It only ran on an Archimedes, so was massively out of date. No one I asked at Acorn knew of it.

Fri, 31 Oct 2025 07:45:43 +0000

So for those of us who are working in C, not C++, and currently using ROOL’s fork of Norcroft, does Piers’ port of the ARM Norcroft offer advantages? For example, is it more up to date with how C has developed? Does it compile code which takes advantage of VFP instead of the old FPA? Or would there need to be updates to other parts of the OS, like the Shared C Library, for any of this to be useful?