Hi Tomasz,

Where should I delete the code that do support this directive in 1.49?
I have done some but i'm afraid it's not that clean.

I have realize that i have no use of it and i want some speed back.

The support for "fix" directive is optimized in a way that there should not be any slow down unless you use this directive at least once in your source.

hmm... I hope so but I got 1.6 sec when compile using 1.49 and 1.1 when compile using 1.47 and the both has been used to compile GUI source file of 1.49 (Fasmw).

I'm on 1.2 GHz Celeron.

Hmmm, that's strange. Try removing line 279 from PREPROCE.INC, it looks like:

hmm.. didn't work.
It seems like there are many place have to be removed, so let's forget about it.

Anyway thank you.

This is a hmm... thread.

Hmmm, what if there is some other problem that makes compilation slower.
I got an impression that 1.49 works slower too, but I have assumed that this is because of bigger Fresh project...

Hi JohnFound,

This problem was introduce sinc 1.48 (the fix directive was born in this version of course).
So I guess it causes and i'm try to think of that this is not a problem.

OK, let's see in the WhatsNew.txt for 1.48:

[+] added fix directive
[-] fixed small bug with symbolic constants

BTW: I ever wonder, what [+] and [-] means. Maybe [-] means slow down of compilator?

hehe.. nice joke

+ sign mean added a feature or something cool to be added.
- sign mean removed a feature (mostly is bug) or get rid of unoptimized code.

Thanks, it was not fully joke. I mean, maybe Privalov, unconsciously make some changes, that slow down FASM. I remember, there was some changes in conjunction with list feature...

[quote="JohnFound"]
Bingo! The slow-down was caused by the new feature of storing the corresponding offsets in output for each source line, which I have added in the way to enable listing/symbolic debugging features. Since official 1.49 does not utilize it anyway, I've removed it - you can grab the updated 1.49 release from the fasm's website.
Additional (much smaller) slowing down is happening for PE output only, and is related to the checksum calculation, which was added recently. If you don't need this feature, you can disable it by commenting out lines 1186-1204 of FORMATS.INC.

Yea, now it is faster, but little. For Fresh project on my job's computer: PII 400MHz:

Was: 10.8s
Now: 9.6s

Regards.

Oche!, good job.