[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [MiNT] AHCC V4.11
On sunnuntai 26 elokuu 2012, Ole wrote:
> did you try to link with ahccstdf.lib?
As you can see from my mail, it needs only the integer version of it,
but I *did* test both.
Of course adding also the non-CF version of the std lib to the project
fixes the issue as that includes the missing function, but I'm
afraid the non-CF versions of mul & ldiv aren't CF compatible...?
(After that project file change AHCC CF version started to bus error though,
and I cannot for some reason get it working anymore although I removed all
intermediate files. :-/)
Depending on link order, AHCC might also include all the rest of the std
functions from that instead from the CF version of std lib also listed
in the project file. I'm assuming AHCC linker takes the function from
the library containing the function that is listed first in the project
> Don't take this as a solution, it's maybe the wrong thing
> to do - even if it works. The library variety which comes
> with AHCC is somewhat difficult and can lead to some problems
> in understanding and the resulting executable.
It's all explained in the AHCC manual.
> Maybe it also makes a difference when you include
> math.h directly
Adding that doesn't seem to help either (I don't use any math
functions or defines so that was expected).
> and also have a look in there - there
> is some switch for 2 different math implementations.
> Only Henk can give a real answer on this. May I tell you that
> he also created an official support thread for AHCC on the
> Also I have a feeling that the latest AHCC wasn't that good tested -
> but that's just my feeling ;)
I completely understand there being bugs in AHCC versions.
Henk is just a single person and I don't think he has a (private)
test suite for it.
There are bugs in nearly all the releases (I've reported Henk
bug(s) in every release I've tested), it's just luck that you
haven't happened to hit them. :-)
Those bugs aren't a problem though as Henk typically responds
very quickly with a fix and so far I've never seen the same
bug, they're always new ones / ones found with some code
(e.g. bug in my own code) I haven't tested earlier.