On May 8, 2004, at 3:06 AM, Ben Hines wrote:



On May 7, 2004, at 4:42 PM, Ben Hines wrote:
That did not happen to me. Perhaps it will only occur with programs which do not use ncurses, but which do use readline? Which programs ?
The 2 cases were from a local program, that linked both readline and ncurses _ in this order.
Clearly the order is important for the trouble _ but you need also that one of the object files does
reference one of those 3 symbols...
A quick sampling (yes _ the script is fast...):
for f in bc cadaver-ssl calc ekg-ssl gftp-ssl ghc guile guile16 hugs librep mutella mysql nickle \
octave-forge octave pilot-link9 python22 python23 r-base rpy-py23 ruby ruby18 samba-ldap \
units yafc yap singular hx term-readline-gnu-pm581 emacs21-xaw3d xemacs bash ; do echo $f; \
dlocate -L $f| xargs file | grep 'Mach\-O' | cut -f 1 -d ':' | xargs nm|grep -F '_BC
_UP
_PC' ; done


yielded only hx in addition (and indeed the command hx breaks as shown yesterday)

Also, at this point we might as well start adding versioned deps on the latest readline on anything this effects and rev upping them.
If we were to do this _ shouldn't we use the opportunity to bring readline up to date with all patches ?
We'll have to do it eventually to solve this. Only about 70 packages depend on readline, and this problem does not affect them all. I've tried a few (older binaries from apt-get) and haven't been able to find even one which does it.

In view of the above test, I'm not convinced this is necessary, as a practical matter.
If there are only about 70 pkgs, it may suffice to run the above test on them...
Of course, if its a matter of (shlibs-) policy, I don't know.. -)


Best,

Jean-Francois



-------------------------------------------------------
This SF.Net email is sponsored by Sleepycat Software
Learn developer strategies Cisco, Motorola, Ericsson & Lucent use to deliver
higher performing products faster, at low TCO.
http://www.sleepycat.com/telcomwpreg.php?From=osdnemail3
_______________________________________________
Fink-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/fink-devel

Reply via email to