| 1 |
On fre, 2003-11-07 at 09:59, Paul de Vrieze wrote: |
| 2 |
[SNIP] |
| 3 |
> > |
| 4 |
> > libmysql links against libssl-0.9.6, and say you only have |
| 5 |
> > openssl-0.9.6 installed at the time. |
| 6 |
> > now you upgrade a slotted openssl, and have both libssl-0.9.6 and |
| 7 |
> > libssl-0.9.7. |
| 8 |
> > now you compile some other program that wants to pull in BOTH libssl |
| 9 |
> > and libmysql. it picks up libssl-0.9.7 as that is the latest, but |
| 10 |
> > libmysql is still linked against libssl-0.9.6. at this point your |
| 11 |
> > program probably won't compile properly (gcc will detect the problem |
| 12 |
> > in most cases), but if it doesn't your program will just segfault when |
| 13 |
> > you try to run it. |
| 14 |
> |
| 15 |
> btw. This is exactly the problem with berkeley db, and it is quite common |
| 16 |
> for berkeley db. |
| 17 |
> |
| 18 |
And for a lot of other programs unfortunately. This is the thing that |
| 19 |
for me is the biggest show-stopper, for server systems - as it requires |
| 20 |
a lot of wellplaced thought and work, before you upgrade f.ex. OpenSSL |
| 21 |
(which has been requireing upgrades quite often recently :( ) |
| 22 |
|
| 23 |
I hope you can find time to look at my rewritten GLEP and see how you |
| 24 |
like it. I've detailed how I think it should be implemented, and it is |
| 25 |
in now way a performance killer, but AFAIK it should catch every |
| 26 |
occurence of this problem. |
| 27 |
-- |
| 28 |
Regards, |
| 29 |
Klavs Klavsen, GSEC - kl@××××.dk - http://www.vsen.dk |
| 30 |
PGP: 7E063C62/2873 188C 968E 600D D8F8 B8DA 3D3A 0B79 7E06 3C62 |
| 31 |
See my new managed CMS Hosting Service at http://www.VirkPaaNettet.dk |
| 32 |
|
| 33 |
Working with Unix is like wrestling a worthy opponent. |
| 34 |
Working with windows is like attacking a small whining child |
| 35 |
who is carrying a .38. |
Archives