A89: Re: Announcement: new version of mlink (kernel support)
[Prev][Next][Index][Thread]
A89: Re: Announcement: new version of mlink (kernel support)
Hi!
> I have added support for kernel-based programs to mlink now. I
> have not done any extensive testing (still beta), but so far it
> seems that the generated program images are (most of the time)
> identical to the ones that Xavier's linker ("link") generates.
> When they differ, it's because of bugs in link, not mlink! :)
> ...
I really want to replace buggy Xavier's linker in the near future
with better linker, but there is one problem: this completely
removes Xavier from the TIGCC project. I think that this is not
fair, because he is in fact the founder of the whole project. Have
anybody any fair idea how to solve this ethical problem?
> "smart nostub" for safe linking of TIGCC programs without the
> need for an external stub
I think that TIGCCLIB must be modified however if mlink becomes
an official TIGCC linker. Fortunately, these modifications may
be done in 1-2 days. For example, I don't believe that a lot of
extensions added in "nostub.h" in tigcclib 2.1 (like SAVE_SCREEN
directive) are compatible with mlink. Or, maybe I am wrong?
By the way, TIGCCLIB 2.2 will be released at Friday. I breaked
the planed release date because I worked hardly on bypassing some
stupid protections which are introduced in AMS 2.04 & AMS 2.05
(which caused second-launch-crash; if you used TI-Chess with
AMS 2.05, you know what I am talking about).
Zeljko
Follow-Ups: