Ada 83 TLALOC on Windows 64/x86_64

Hi.all

In one intense evening with AI help The assembly of TLALOC LLIR has been ported to Windows 64 and tested under wine. There is a codi_x86_64W.finc (with a “W”). which is used with fasmg to assemble the FAS/FINCs in a PE64 executable.

First the compile/assembly is made on Linux, the cross PE64 is then used with wine to recompile everything and produce a cross compiler x86_64exe for Linux from Windows (follow me ?)

But producing a Windows executable from Windows must be done via fasmg because TARGET_CODE to another OS than Linux is not currently done. It is just a transposition of the SYS_ services of the new codi, it should not be very complicated.

In fact changing the target OS is relatively easy because only 15 SYS_ macros must be adapted, all other binary generating macros do not change.

The LLIR macros is a very practical intermediate, the front end always produces the same LLIR for all processors and all OSes. Then the LLIR stack machine macros are adapted to the targets.

As soon as possible, I’ll try to make some optimization because the execs are big.

3 Likes

Sorry, I don’t understand any of this vernacular.
IIUC, there is TLALOC on Wndows now. Where can I get it, how to install and operate it?
Ececs may be big - i don’t care, just for playing a bit with it.
Christoph

Hi Christoph.

I prepared a cut down to bare minimum repos for you to try. Have a look at :

The README gives simple instructions, tell me where it goes wrong if it does. I can only test with wine on Linux, I have no Windows 64 at hand. Access to sources remains on the other repo :

The assembler link is picking up the closing parenthesis; easily fixable by a user, but thought you’d like to know about the broken link.

Hi OWS ! Thank you for the return ! corrected.

I had some free time and ran through the steps.
Everything compiled and Bonjour ran as expected.

Looks good, insofar as I can tell.

1 Like

Hi OWS.
Nice ! thank you for the return.

Not sure if you were aware, but SourceHut has recently changed their policies to no longer allow projects that use LLMs.

Hi csb6 !

Thank you for the information, I did not know about this. I’ll migrate everything to Github. I am not a very heavy user of AI in the sense that I have no automated process on repositories. I only use them as backup of my own computer. But source hut position is defendable though some project as TLALOC would never have progressed without AI. I hope and myself thinks about open source and lower energy demanding AI. Well used AI is nonetheless a valuable tool.