Since the 1.61.x development line already has all the features I planned for the 1.62, the stable version is going to be released soon. For the purpose of Win64 programming the "MS64 COFF" output format is implemented and tested to work with the Microsoft's linker for Win64. The direct production of PE+ executables for Win64 is planned for the stable 1.64 release on my roadmap - this is why I said once that the 1.64 is going to be the final release for x86-64; however the x86-64 support is in general already finished and stable (there were still some bugs in ELF64 object output, but got fixed during the 1.61.x development).

The biggest changes were applied to preprocessor (see http://board.flatassembler.net/topic.php?t=3592), and the large update of documentation had to be done due to them. Not only the new sections about added features, but also some corrections and additions to other sections, explaining some important details I forgot to document earlier. For example finally the symbol characters recognized by fasm are listed in the section 1.2.1 (and terminology related to tokenization of lines done by preprocessor is explained, these definitions are important to understand many details of the macroinstruction processing).

Do you consider extending macro library in future versions, for example adding redefined directives such as "db" (as you wrote here)?
After introducing new features, a lot of interesting solution appeared, and some of them could be implemented in official package (not only extended .if macro).

Yes, but some of the new solutions, even though much better, would be incompatible with the older macros. There's a possibility that completely new macro package will be designed, and while distributing this new package and examples based on it, still distribute also the old macro packages for compatibility.

I think the compatability issue at the moment is not a problem, isn't it?
The developement of Fasm though goes very far by now but it is still young and not many code for its around (compared to large Tasm and Masm bases).

So I think it would be a good idea to make change by now or otherwise we could step into problem like Masm. It is my personal opinion though.

yeah, rather make new macros. Do we want to end up like all that backward-compatible-overbloated stuff?

I understand now why MenuetOS kernel source code rarely use macros - there will be no problem when Privalov break backward compatibility. There will be no problems with applications if macro behaviour willn't changed (MenuetOS applications don't use standart .inc files).

The standard includes apply only to Win32 programs currently - there are no official macro packages for DOS or Linux in fasm distributions.

I have 1 more feature request for fasm: couldn't you add support for AMD's extended 3D-Now! instructions? Though this shouldn't be too hard to add, since fasm already supports basic 3DNow-instructions. It's strange, I know more PC's here that support the extended 3D-Now! instructions than SSE2/3, since they are all Athlons/XP+/64 rather than Pentium 4s/Ms.

The extended 3DNow! instructions got all already implemented quite a few versions ago, and are documented just among the basic 3DNow! instructions in the fasm's manual.

Oh, it has just reminded me: I've got to document the newly implemented VMX instructions, too.

Ahhh, maybe I just overread them all, or I didn't distinguished them from the basic 3D-Now! instructions because they don't have an own chapter number in the docs

I have released the 1.62 today and updated the documentation and examples on the website aswell.

For anyone wanting to write any macros I highly recommend read the updated chapter about preprocessor in the manual, for instance the new section 2.3.7 contains some important informations about preprocessor which were not really documented till today.

Just wondered if it's necessary with both "repeat ... end repeat" and "rept { ... }"? Or can I accomplish the same with rept as with repeat - if so, I find the old "repeat ... end repeat"-block superfluous... Comments?

Look for example at XOR Crypting macro - generally "repeat" is needed where the repetitions process is depended on some values resolved at assembly time.

I see...

I have decided to release the 1.62 even though I haven't finished all the additions to the manual I planned (the new subsection in 2.1 about the VMX instructions, and also the new subsection in 2.2 about code resolving, the issues which are already mentioned in the Design Principles article, and also about combining the control directives to get interesting results). This is because it's likely that I won't have as much time for fasm in the next few weeks, and I didn't want to delay the new release for this reason - it defines some new standards so the sooner it was released, the better.

I have just got back to my desk after a few days and see that a new version has already been released.

So much of my code has now been broken with the new version

I was able to do a small amount of editing in my standard includes (mostly with 'fix') but I still have one problem that is causing most of my code to be broken:



At this point I have gone back to 1.61.9 so that I can continue to do some coding without having to make major changes to all my previous code.

Can the new macros be altered to allow the above to assemble? I don't yet understand them so I haven't been able to make a correction yet.

I also have a problem with the new version of the struct macro, the 'times' directive doesn't work correctly. Is there a way to make it work correctly?

One other thing I forgot to mention.

With the new struct macros I have to be careful when using this:




And also one more thing I noticed. When an error is detected in an "included" file, the error shown does not show the "path" to the included file. That is, it does not show the line in the source where the file was included.

I made a mistake once of including the wrong file name and later when the error came up I had trouble finding the original line with the "include" directive.

First of all, this is the place to discuss the new macros:
http://board.flatassembler.net/topic.php?t=3619
(I have just made it sticky so it will be harder to miss it)
It explains how the new macros should be used.

madmatt: instead of times you should use the "dup" operator with standard data directives, only the data directives are allowed inside the "struct" (it was designed to imitate the syntax of TASM/MASM, for the fasm's syntax just go on with the plain "struc").

revolution: making the include traceable like macros are is not possible with current design of preprocessor's data structures, so I'm sorry, adding such feature won't come easy. Until your comment I've never realized that such thing may come useful.

The "x equ dqword" problem you mentioned was actually the bug in fasm itself, I've fixed it now and updated the 1.62 with the fix. Yeah, I knew that there was not enough testing of 1.61.x... :/ The long tradition of hidden fixes is kept.

The attached file contains the old "proc" macros updated to work with fasm 1.62 - you can use them to compile the sources that use the old syntax.

Or even, if you use the old syntax and have placed "enter" in each procedure, you can use such wrappers over the new macros:

All Right, thanks.