Thanks for the quick reply! :)
Julien Pierre wrote:
> You shouldn't be defining things like i386 since you are not targeting > an i386 architecture. >
Hmm - I didn't know I was! :) This output was generated on the PlayStation 2 Linux environment, with a standard .mozconfig - basically the same suggested by Mozilla.org for release builds. (I can provide if necessary.)
Christopher Seawood pointed out in a message when I was enquiring on netscape.public.mozilla.unix that "If you look at security/coreconf/Linux.mk, it defaults to x86 if it cannot find a supported arch. You'll need to file a bug on the NSS product and probably post to m.crypto to ask for help in porting to the mipsEEel arch."
That's what led me here! :)
There is a C implementation of the mpi library that you can use. It should be used on any platform that we haven't explicitly provided assembly code for. It will of course be a little slower than assembly, but that should get your build going. The performance difference is most visible in server applications rather than client applications, so you may be able to live with it even for the release build.
This sounds fine by me. How do I build Mozilla 1.3 with this? Is there a webpage with instructions, etc. that I can follow? Sorry - I'm mostly a Mozilla builder - not so much a programmer.
Also - most of the time I build with a cross-compiler. Are there extra steps to get NSS to build on a cross-compiler?
FYI, on OS/2 we didn't use the assembly code for years in the browser buil, until I found out and fixed it. See http://bugzilla.mozilla.org/show_bug.cgi?id=104740 . This bug has pointers to the files I fixed to get the assembly code linking on OS/2. Of course that was an x86 platform so the changes were minimal.
Thanks for pointing this out - I'll definitely take a look at it.
Cheers, Paul
P.S. remove nospam. from my reply-to address to e-mail me directly
