Hi, All!
Can anybody explain mesuch thing. I am writing new OS and want to port FASM there. Can i do this (i mean rights for compiler) - if i will keep license file etc, but i need to rewrite some parts of code.
Is it legal to do so?

Nobody can answer such simple question? Please...

If you mean parts like SYSTEM.INC or FASM.ASM then it's okay As for platform-independent part of assembler, ask someone else

You can modify it and distribute modified versions as long as you follow the restrictions detailed in the license.txt that you get when you download fasm.

To port fasm to your os you have to modify system.inc and fasm.asm (as mike.dld wrote). These files provide an interface to an operating system. You have to put OS specific routines (reading from files, displaying text, etc...) in system.inc file. It would be easier if you copy these two files from win32 distribution and modify them.

Thank you!

The attachments contains the copy of the never finished guide to the flat assembler internals, slightly updated to cover the latest releases.
The second chapter is however complete, and this is the one that should help you in porting fasm to any new operating system.

Is this attachment the most newer version of the guide?

Unless there is another version outside this forum, this one is the last release of the guide.

Thomas, can u find some spare time to update this fasm1 guide ? Would be much much much appreciated.. it's almost 14years old now.. and i want to tinker with fasm engine.. unfort lots of information are missing in there : typ. description of parsing, assembly, formatting stages (am particularly interested to see how virtual feature work internally, how it could be extended to include relocations in the case of PE format)
Thx
Rgds

This old engine is kind of a mess, its internal structures have been expanded so many times (including recent back-porting of VIRTUAL-related features from fasmg) that it became a bit like a patchwork. It is not easy to convince me to get back to analyzing it when I have a much more cleaner world of fasmg to spend my time in.

On the other hand, I still own a notebook where I designed all the data structures used internally by fasm 1. I might consider digitizing it one day... Several structures have already been described in the fas format documentation.

If by digitising it you mean to replace all the raw numbers with symbolic constants that would be a significant improvement.

I attempted this several times in the past (not recently, though). Each time I had to abort after realizing that I've been using some of the values with multiple meanings. You know, like Mel would do.

Nevertheless, in the fasmg development I have in fact found a good way to handle this kind of problems - with the ASSERT directive. When I'm going to use some offset or a constant in such way that changing the original definition of structure or constant would break things, I simply add an ASSERT that is going to signalize the problem. There are many such clauses throughout fasmg source.

So yeah, with this new approach I could most likely succeed in rewriting fasm's source in such way if I attempted it. The only problem now is, as I wrote above, to somehow make me work on fasm 1 source when I so much prefer to dedicate the same time to fasmg-related things.

Maybe there could be a guide on how Tomasz can migrate from fasmg to fasm. You would experience much faster assembly times, so testing is much easier and faster.