Bug 193298 - Libbind patch in squid SRPM fails
Summary: Libbind patch in squid SRPM fails
Keywords:
Status: CLOSED RAWHIDE
Alias: None
Product: Fedora
Classification: Fedora
Component: squid
Version: 5
Hardware: All
OS: Linux
medium
medium
Target Milestone: ---
Assignee: Martin Stransky
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2006-05-26 22:12 UTC by John Horne
Modified: 2007-11-30 22:11 UTC (History)
0 users

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2006-05-28 11:42:31 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)

Description John Horne 2006-05-26 22:12:52 UTC
Description of problem:
The squid-2.5.STABLE12-libbind.patch in the current squid SRPM causes
an error during the rebuild.

Version-Release number of selected component (if applicable):
Squid version squid-2.5.STABLE13-1

How reproducible:
Every time.

Steps to Reproduce:
1. Download and install lastest version of FC5 squid SRPM
2. Enter commands:
     cd /usr/src/redhat/SPECS
     rpmbuild -ba squid.spec >/tmp/out 2>&1
3.
  
Actual results:
View output file /tmp/out and it will show:

     checking for main in -lregex... no
     ./configure: line 7195: test: =: unary operator expected
     checking for main in -lm... yes

Expected results:
No error should be generated.

Additional info:

The patch removes the configure test for gethostbyname in lbind by commenting
it all out. However, as the last 2 lines of the patch shows:

    if test $ac_cv_lib_bind_gethostbyname= "no" ; then
     case "$host" in

Because all the test is commented the variable '$ac_cv_lib_bind_gethostbyname'
is never set - hence the test is testing something undefined against "no" and so
causes the error.

I suspect because of this error the subsequent tests of lresolv in configure are
skipped as well. This makes me wonder if the patch is needed; squid seems to
work fine despite the error.


John.

Comment 1 John Horne 2006-05-26 22:56:05 UTC
1) This bug also exists under FC4.

2) By modifying the last commented-out line of the patch from:

       +#fi
   to
       +ac_cv_lib_bind_gethostbyname=no

   the patch now works correctly, and the output file (/tmp/out) shows that
   additional checks are being made (this was under FC4):

       checking for main in -lregex... no
       checking for inet_aton in -lresolv... yes
       checking for inet_aton in -l44bsd... no
       checking for main in -lresolv... yes
       checking for main in -lm... yes

   The middle 3 lines are new - because the patch now works.

John.


Comment 2 Martin Stransky 2006-05-28 11:42:31 UTC
Fixed in rawhide, thanks.


Note You need to log in before you can comment on or make changes to this bug.