Re: [MACTCL] Bug in 8.4.10 tclLoadDyld.c?

[Daniel A. S. <st...@ic...>]

Jeff,

On 29/07/2005, at 5:02, Jeff Hobbs wrote:

> The problem does lie in TCL_LOAD_FROM_MEMORY, but I'm not
> certain that it is particularly any code at fault there, or
> if that is simply not compatible with metakit and its
> memory channel stuff.

I don't think that's should cause any issue, all we are doing for 
TCL_LOAD_FROM_MEMORY is reading from a file in a VFS with standard tcl 
io... and it works with .bundles anyway, I've tested that (and e.g. the 
tk included in tclkit is linked as a .bundle and loads just fine from 
memory).

The problem must come in the fallthrough to standard behaviour of 
copying out to temp location later on (the load from memory code tries 
and fails by design on .dylib's)

in any case, it appears that this problem is on Tiger only, steve's 
test starkit works fine for me under tclkit on Panther even when 
wrapped. can't test/debug on Tiger ATM, but will try to look into it 
over the weekend.

> In any case, I have commited a fix for 8.4 that prevents
> that from being defined in the static build case.  Standard
> builds will still use it, but basekit builds won't.  I have
> confirmed that this works.

ok, that's not a good permanent fix though since a standard tclsh can 
also read starkits with the appropriate extensions, so the problem will 
appear there as well on Tiger.

Apple made significant changes to dyld in Tiger, they may well have 
broken something...

Cheers,

Daniel

-- 
** Daniel A. Steffen     **  "And now for something completely
** Dept. of Mathematics  **   different"    Monty Python
** Macquarie University  **  <mailto:st...@ma...>
** NSW 2109 Australia    **  <http://www.maths.mq.edu.au/~steffen/>