| 1 |
Neil Bothwick <neil@××××××××××.uk> posted |
| 2 |
20061017084723.500c31ca@××××××××××××××××××.uk, excerpted below, on Tue, |
| 3 |
17 Oct 2006 08:47:23 +0100: |
| 4 |
|
| 5 |
> On Mon, 16 Oct 2006 23:33:06 -0700 (PDT), Steve Herber wrote: |
| 6 |
> |
| 7 |
>> configure: error: There is something wrong. Please check config.log for |
| 8 |
>> more information. |
| 9 |
>> |
| 10 |
>> !!! ERROR: dev-lang/php-5.1.6-r6 failed. |
| 11 |
> |
| 12 |
>> Some of my major packages get this error and I wonder what causes it? |
| 13 |
>> I am not interested in a solution for php, but a guideline to help me |
| 14 |
>> fix any package with this problem. |
| 15 |
> |
| 16 |
> The lines before the error should give some clues. The config.log it |
| 17 |
> mentions can be found in $PORT_TMPDIR/dev-lang/php-5.1.6/work in this |
| 18 |
> instance, that should give a more detailed error message. |
| 19 |
> |
| 20 |
> This isn't necessarily a single problem for all the affected packages; the |
| 21 |
> configure script carries out a large number of tests, any one of which |
| 22 |
> could have failed. |
| 23 |
|
| 24 |
Additionally, note that with some tests, any warnings (even if they are |
| 25 |
harmless) can cause the test to fail. A common example would be |
| 26 |
unsupported CFLAGS -- CFLAGS that your currently configured version of GCC |
| 27 |
doesn't recognize and so ignores, but spits a warning about. Maybe you |
| 28 |
setup your CFLAGS with a different version of GCC that recognized |
| 29 |
different flags, or whatever you followed suggesting those CFLAGS was |
| 30 |
based on a different GCC version. Particularly if whatever they are |
| 31 |
testing is only a gcc-warning as well, guess what, they usually fail on |
| 32 |
any warning. As it happens, the Gentoo/amd64 arch team had enough problems |
| 33 |
reported due to this that they instituted a filter in their profiles that |
| 34 |
removes most flags the currently configured version of gcc doesn't |
| 35 |
recognize, so the problem happens less frequently than it did, but the |
| 36 |
unrecognized CFLAG warning isn't the only one gcc has by far, and if one |
| 37 |
of the others pops up in a test, it could trigger a test failure and |
| 38 |
ultimately a configure error as well. |
| 39 |
|
| 40 |
In ordered to properly troubleshoot the general problem (a configure error |
| 41 |
as reported in config.log), one really needs to know bash (well sh, which |
| 42 |
on Gentoo is normally bash) scripting, since that's what the configure |
| 43 |
script is written in. The config.log file references line numbers in the |
| 44 |
actual configure script, and if it's not immediately apparent from |
| 45 |
config.log what the problem is, the next step is to open the configure |
| 46 |
script to that line and figure out what it's doing, duplicating the |
| 47 |
process by hand or adding echo debugs to the configure script where |
| 48 |
necessary to see what's going on. With a decent grasp of bash and at |
| 49 |
least a vague idea of what all those tests in the configure script are all |
| 50 |
about (note, knowing C/C++ is generally NOT necessary), one can usually |
| 51 |
figure out what's going on and why, either correcting the problem (such as |
| 52 |
CFLAGS that are choking the test in question) or bugging either with |
| 53 |
Gentoo or very occasionally upstream, as appropriate. |
| 54 |
|
| 55 |
-- |
| 56 |
Duncan - List replies preferred. No HTML msgs. |
| 57 |
"Every nonfree program has a lord, a master -- |
| 58 |
and if you use the program, he is your master." Richard Stallman |
| 59 |
|
| 60 |
-- |
| 61 |
gentoo-amd64@g.o mailing list |
Archives