On Tue, 11 Dec 2001 17:22:01 +0100, [EMAIL PROTECTED] wrote:
>I prefer VP too, because of its good native performance and because it
>generate fast and compact executables.
First of all, we must have some words about future of vp. If development will be
stopped, then can we get sources of compiler and rtl? Can we distribute hwem?
All other words about vp unrequired. I agree, vp greatest compiler at this time.
>FPC is a choice, but not IMHO a good one, because of little support for
>OS/2 and because of EMX. Please note that EMX is not "only a library",
>rather it is a layer aimed at providing some unix compatibility to OS/2.
>Given this origin, EMX implements many functions in a less than optimal
>way, sacrificing performance for compatibility. For example, the FPC
>library probably contains support for Posix threads, NOT for OS/2 native
>threads, so we would have to rewrite (or provide a wrapper for) that
>part of the RTL. We could compile it with EMX and obtain working
>threads, but it would not be a good implementation. Another example is
>the wide use of "fork" in unix, which is implemented in OS/2 via EMX,
>but provides poor performance and poor memory optimization.
>I think that in the end having a good RTL for FPC in OS/2 would be MUCH
>work, maybe much more than adapting other native RTLs.
Sorry, but this correct for GCC+EMX, but not for FPC/EMX. emxlibc uses fork, posix
threads, etc. But fpc rtl seems to not use this. viocalls unit uses emx dlls, but pm*
uses
pm*.dll without any wrapper. At this point emx just a library. All words correct
about ported *nix software, but for native software we can and must use native api,
not emx
wrappers (exclude 16 bit functions, but is 16bit api calls required???). I agree, that
fpc doesn't have fully implemented os/2 api calls. I think fpc can be a choice,
because it
is open source and we never lost compiler but with vp we can. I agrre, that moving to
fpc from vp easely, whan sibyl->vp/fpc. And easely port sibyl->vp.
>As a last note, Allan Mertner (VP author) once showed support for the
>idea of porting SPCC to VP, and told that he would have made changes to
>the compiler if required.
>
>I don't know if he still thinks so, but asking doesn't hurts.
>I think that if we write to [EMAIL PROTECTED], the message will go to
>Allan (Adrian? Will you try?).
Another way. We can port to vp and extend support of os/2 in fpc. Can Allan contribute
some sources of rtl to fpc team? I don't know. If can - good.
Again. I agree with vp choice. Its great compiler. But how about future??? I don't
want make hard work and lost big part of them in future. Only this can stop me. Only
because fpc can be good choice.
Thats all.
-----------
To unsubscribe yourself from this list, send the following message
to [EMAIL PROTECTED]
unsubscribe sibyl
end
