Notes, Change log, etc. for Att2Intl



Still to do...
  * There are probably some bugs that need to be worked out. The section 
    starting around line 746, for instance, may not insert []s in some cases
    where they may be needed in NASM output, although this has now been 
    greatly improved and may in fact be completely right.
  * Maybe there should be a better way of figuring out which simbols need
    extern declarations. Calling tasm or nasm is crufty even though it works.
  * Since NASM exists on FreeBSD, Linux, etc. and not just DOS, maybe I should
    make sure that att2intl works there.
  * For my own purposes I will add the option of handling segments differently
    in NASM output. 



Code contributed by...

Gabor Keresztfalvi, keresztg@podolin.piar.hu, http://www.piar.hy/~kerestg/




Change log...

v0.2.1 -  Aug 20, 1998

Gabor Keresztfalvi writes:
  - Better []s use recognition (thanx Hmaon :)
  - Fixed short jump error with multiple running of Nasm
        Geee. This was a stupid bug...
  - New parameter: -b use BSS when found .comm or .lcomm
        This is not the best yet, because I suppose that all of the .comm and
        .lcomm come in the end of the AT&T source...
  - group definition DGROUP if needed
  - if debug level is 0, then cleans the temporary files
  - Fixed multiple use of .text and .data in the AT&T source


v0.2.0 -  Aug 16, 1998

Many thanks to Gabor Keresztfalvi, who implemented the following improvements:
  * NASM compilable output. att2intl will now output NASM compilable code 
    by default. You can switch to TASM output with a command line switch.
  * Debug output is now controlled with a command line paramemter. Debug
    output is suppressed by default, use -d 2 to see it.
  * additional command line parameters, ie -r
  * multiple extern declarations are now filtered out so each symbol only
    appears once


v0.1.6 -  Apr 6,  1998

Bugfix:
  * in checking for external symbols (by calling tasm), att2intl now knows what
    to do if it encounters a "Too many warnings," or whatever it says, message
  * correctly handles a mul or imul with three paremeters, 
    ie "imul $80,%ebx,%edx"
    Actually, anything with three parameters will be translated quitely
    by reversing the parameters


v0.1.5 -  Feb 13, 1998

Improvement:
  * C style comments allowed in source file. (added by request)


v0.1.4 -  Jan 30, 1998

Bugfix:
  * now parses lines such as "call *%eax" correctly. I don't really know why
    gcc puts a '*' there, I don't remember info having any info about it.
  * aligns are correctly converted, or maybe that was fixed in an earlier
    version. Anyway, they're correct now, that's for sure.


v0.1.3 -  Jan 17, 1998

Bugfix:
  * fixed stray DWORD PTR
  * fixed spelling of the word 'existence'


v0.1.2 -  Jan 13, 1998

Bugfixes, bugfixes:
  * \n, which gcc translates to \12, is expanded to 13, 10 as opposed to
    just 10
  * Tasm compiles "push DWORD PTR (offset Label)" to "push DWORD PTR [Label]"
    sometimes it even compiles variations of the command to
    push bp
    mov bp, sp
    ;some junk
    pop bp
    Sheesh... I changed it so it doesn't output a DWORD PTR for push
    but you need to double check the output if you want to generate 16 bit code
    well, like I said, nasm support coming soon.


v0.1.1 -  Jan 12, 1998

Improvements:
  * program now cleans up after itself by deleting its temporary files

Bugfixes:
  * fnldcw and related instructions handled correctly, the w is no longer
    hacked off
  * argument size translated correctly for instructions starting with 'fi'
    ie fisub and fild
  * 'jl' no longer truncated to 'j'. Never code if you're really tired.


v0.1 -  Jan 11, 1998

First version

Improvements:
  * existence

Bugfixes:
  * apparently needed more

--
Hmaon / Xylem
Greg Velichansky
www.wam.umd.edu/~gvelicha/