AsmPro 1.19

[a/b]

Aminet, same place as 1.18:
http://aminet.net/dev/asm/ASMPro1.20.lha
http://aminet.net/dev/asm/ASMPro1.20src.lha
That's for 1.20b.
1.20c (or d, ...) is still work in progress, once there's more bugfixes and/or improvements I'll release a new version. The one I put in the zone for 8bitbubsy is basically 1.20b + optional multiline scroll when assembling.

[hitchhikr]

I've seen that one but it seems to be old.
I'll wait then.

[malko]

Originally Posted by Photon

[...] I'm still finding time to finish the todo-list for a finished AsmTwo [...]

!
Keep up the good work

[kamelito]

@a/b
I tried to assemble a source, the source include graphics/gel.i at line 138 of this include there is : AC EQU 0
Asmpro (and probably one) trigger the error ; ** Reserved word
I can understand that ac is a command in the assembler shell but why complain about it in the source code?
I got the same thing because in the source there is a label named : tt1, dunno what is the purpose of tt1 in that case I just changed the label name but I can’t do that in official includes.
There is maybe other cases…

[a/b]

Those are MMU registers.

[kamelito]

So because Asmpro set MMU registers as reserved keywords a developer cannot use gel.i include in his code and label like tt1 and probably others ?
This is not a problem with Devpac for example and I have MMU enabled in the settings.

[a/b]

That's how asm-xxx are designed. You can use any register even with 68000 selected and it will assemble, but you can optionally enable warnings to see possible problems. Consequently, all register names of all units are reserved words. Changing that would require substantial changes.
Devpac obviously handles that differently, and it works in *practical* cases (you can still construct code that will cause it to fail or produce unexpected opcodes, but you can control that via your source code so it's fixable).
I don't know how to handle this...

If I look at what's going on in the includes, the problem is "STRUCTURE AC,0" (name, initial offset), which is translated via macro (exec/types.i) to "AC EQU 0" and "SOFFSET EQU \2".
Now, I don't know why exactly is the first EQU needed at all, especially when always set to a zero. For a possible e.g. IFD/IFND at a later time to determine whether that structure is defined? I'm not that familiar with all the intricacies of the includes, so I'd guess it's not needed. I also did a "grep AC" on all the includes and couldn't see anything that would resemble that symbol being used.

So I can only suggest an unpopular workaround: change the include to "STRUCTURE AComp,0".

[kamelito]

thanks

[hitchhikr]

Btw, try to load a binary and input a beginning address like "d$40000".

[a/b]

Originally Posted by hitchhikr

Btw, try to load a binary and input a beginning address like "d$40000".

The usual case of non-sanitized input ;p, and there is/was more of that. You don't realize how borked (well, to be fair, short+quick seemed to be a common theme) some things are until you go through the code. Or you're theorizing/writing something similar and ask yourself, how does asm-xxx handle something that's off the beaten path, and you go do it and realize it doesn't.
Thanks for reporting, I'll have a look right away. I'm surprised this hasn't been noticed before, since a too quick arrow-up/down a few times + enter at the wrong time shouldn't be that rare (this is an old bug, 1.18 or earlier).

[hitchhikr]

I don't know when this appeared because even the first versions of AsmOne didn't crash on this and properly reported an unknown symbol output.

[a/b]

I don't know when exactly it was introduced, but the code is different from Rune's asm-one v1.x. Basically, at some point the read begin/end address (INPUTBEGINEND) routine was expanded with additional d7 parameter to signal optional/mandatory end address, all while lower level parser routines are using d7 as AF_xxx flags, thus trashing them. So if a parse error occurs, the error handler has wrong flags and starts nuking zero page (null ptr). All the other stuff is sending d7=0, only readbin is sending d7=-1.