Re: [squid-users] Solaris/OpenSSL/MD5 Issues

From: Amos Jeffries <squid3@dont-contact.us>
Date: Sun, 11 Nov 2007 23:24:43 +1300

Randall DuCharme wrote:
>>> Undefined first referenced
>>> symbol in file
>>> esiEnableProcessing(HttpReply*) client_side_reply.o
>>> esiProcessStream(clientStreamNode*, ClientHttpRequest*, HttpReply*, StoreIOBuffer) client_side_reply.o
>
>> What does adding a
>> #if ESI
>> #error Something defined ESI!
>> #endif
>
>> Do to the compile process if its added near the top of
>> src/client_side_reply.cc It may be a mater of making ESI use a
>> different bounding macro (USE_ESI).
>
>> Amos
>
> Nothing. I never hits the error. Bizarre. Just as a hack I tried
> commenting out the blocks of code inside the #if ESI/#endif sections
> of the affected modules and was able to get a binary. It crashes
> every time it's accessed accessed however though that may be yet
> another issue.
>

Well, the error was meant to prove whether or not the esi* function
calls in client_side_reply.cc were even compiled. Apparently not. Yet
something is pulling the symbols into the linker.

Where else outside client_side_reply.cc did you remove #if ESI / #endif
code? you said 'modules' so I am assuming the problem came up elsewhere
later.

After Guidos comment, I think we should do with a diff of a good working
include/autoconf.h against yours and see whats up.

Guido: are you able to supply the working version please?

Amos
Received on Sun Nov 11 2007 - 03:24:41 MST

This archive was generated by hypermail pre-2.1.9 : Sat Dec 01 2007 - 12:00:05 MST