Fink on Darwin sets ld environment variables

Fink on Darwin sets ld environment variables

[Brad Lucier]

There was a recent question about link failures in the f77 test suite on
MacOS due to extra warnings.  These happen because fink 6.2 defines several
linker-related environment variables; here's part of the output of printenv
on my box:

LD_SEG_ADDR_TABLE=/sw/var/lib/fink/prebound/seg_addr_table
LD_PREBIND=1
LD_PREBIND_ALLOW_OVERLAP=1

This will result in warnings from the linker if you happen to specify options
that override these environment variables.  (The one that bit me was LD_PREBIND)


Brad

Re: Fink on Darwin sets ld environment variables

[Andreas Tobler]

Jack,

Jack Howarth wrote:
> Brad,
>    Thanks for figuring that out. However I still don't quite understand
> why these link errors appear only in the newer snapshots and the older 
> 20031015 snapshot doesn't show them. 

Most probably due to this one :)

http://gcc.gnu.org/ml/gcc-patches/2003-11/msg01441.html

> Also, how did you suppress them
> for the building of current gcc 3.4 g77? Lastly, I am also a bit 
> concerned about what breakage this issue will cause us...
> 
> http://gcc.gnu.org/ml/gcc/2003-11/msg01346.html

Don't know if this one and the related ones have an impact. I could be 
wrong, but I don't see any similar warnings here. Well, I DON'T use fink.

> 
> since it only appeared in earlier November and apparently still
> exists in the main trunk.

You may try to build your chain with --disable-shared. But then, in case 
you build libgcj, this one will not work.

Personally I find it a bit strange to set linker flags in the 
environment. But this is my personal view.


Andreas

re: Fink on Darwin sets ld environment variables

[Jack Howarth]

Brad,
   Thanks for figuring that out. However I still don't quite understand
why these link errors appear only in the newer snapshots and the older 
20031015 snapshot doesn't show them. Also, how did you suppress them
for the building of current gcc 3.4 g77? Lastly, I am also a bit 
concerned about what breakage this issue will cause us...

http://gcc.gnu.org/ml/gcc/2003-11/msg01346.html

since it only appeared in earlier November and apparently still
exists in the main trunk.
                         Jack