in 1.67.24, there was still an error of invalid parameter when more than one parameter is not defined, possible they are linked?

I find version 1.67.24 to be very sensitive to previous programs. If a certain programs (I think buggy ones that cause memory corruption) have run before it, it does not compile successfully (like generating errors on valid source files, ones that compile after reboot). I found 2 programs (help.com, which seems to just call help.exe (help.exe is fine, help.com caused these problems); and a DOS version of the UNIX command ls) that don't work with 1.67.24 (both weren't very useful, so I deleted them). Still, it works fine under Windows XP DOS box, and I haven't had any problems with the Windows version. Older versions of fasm seem to run just fine even with ls and help.com run before them.

EDIT: I don't know of any other programs that had problems with ls or help.com. Repeated tests showed 1.67.24 to work until I ran these programs, after which it invariably stopped working.

This bug is really weird. Some of the programs I made, even though I know they are bug free, mess up fasm. Anyway, I got the bug to work (for 1.67.24) under WinXP DOS box (see the attached image, it's the DOS version under NTVDM, NOT the windows version). See the change of the memory report, just by running fasm?

The IDE version (and Windows version) don't have this bug. Maybe someone should just take the IDE version, turn it into the cmd version, and release the new version with the cmd version good as well. The bug only occurs under DOS (if it existed under Windows, someone would have noticed). This means the bug is in the DOS directory. I've been using an older version (it works like a charm) for a long time, and I'll use that version until the newer versions are more stable.[/img]

EDIT: removed screenshot, bug fixed

Anyone notice that 256 is correctly out of range for a 8bit register?

8bit registers can only hold 0 - 255?

exact, it's the default, fasm crash on this error. and that's why we test it with 256, to obtain the error..

These issues should be (mostly) fixed by now.

See here.

> Anyone notice that 256 is correctly out of range for a 8bit register?

Did you notice, that we are exploiting bugs ? The "FASMW" whining is good, the "FASMD" evil.

> These issues should be (mostly) fixed by now.

Yeah ... seems to be What was the bug ?

I guess just lack of time for testing and more complexity due to adding "-d".

I've got good news for you! It seems that the problem is solved, because fasm 1.67.31
seems to work properly. Althouh version 1.67.23- have the same problem.

Previously I had to stick with 1.67.22, but now I was able to get the newest
version .). Maybe it was related to problems reported by the others and was fixed along those fixes?

regards
Mac2004

IIRC it's fixed since 1.67.26



Then you missed important new features and bugfixes.

Next time please report bugs in FASM DOS version, rather than silently waiting for an accidental fix for years. 1.67.26 and 1.67.29 work very well for me, although I use the IDE almost exclusively so I might miss bugs specific to the non-IDE version as I did with 1.67.25.

Also, if you indeed have M.E. then Win32 version might be the better choice for you

At least I did report the problem, but Thomazs couldn't reproduce the same bug.

Yes. I'am happy to be able to use the latest version.

Regards
Mac2004