in reply to Re^4: Inline::C produces compilation errors
in thread Inline::C produces compilation errors
|
|---|
| Replies are listed 'Best First'. | |
|---|---|
|
Re^6: Inline::C produces compilation errors
by holandes777 (Scribe) on Nov 24, 2007 at 14:06 UTC | |
I reinstalled version-0.74 and Parse-RecDescent-v1.95.1, both of whom are prerequesistes to installing Inline::C. The make test ended with "Failed 5/6 test scripts, 16.67% okay. 8/11 subtests failed, 27.27% okay." as it had before. When I tried running the original program I get the same result.So I tried the following minimalist program: And I got the following output: So I can compile something, I then kept adding the lines I needed one at a time until I got to this: Which broke complaining that :perl: symbol lookup error: /HASP/perltestP/_Inline/lib/auto/c2_pl_4339/c2_pl_4339.so: undefined symbol: hasp_login". However, if I approach this from the C side as follows, it recolves hasp_login without a problem. This would indicate that Inline::C is unable to resolve extenal references that are not from itself ??? I think it all relates the the tests that are being failed on the "make test" of the Inlince::C installation. I've googled and I have checked the bug reports in cpan.org. Nothing really looks like this. Should I report it as a bug to "INGY" on cpan.org's site? | [reply] [d/l] [select] |
by syphilis (Archbishop) on Nov 24, 2007 at 22:29 UTC | |
I think that's wrong, unless the library is called something like libhasp_linux.a.a. Or is the linker smart enough to know that the '.a' should be dropped ? I would change it to: and check the output for any messages that tell you that no library was found for -llibhasp_linux. When you build the C program do you provide an -llibhasp_linux.a argument on the command line (or is it simply -llibhasp_linux ) ? I would expect that whatever argument works on the command line should also work for the 'LIBS' config option. Note: The BUILD_NOISY option means that you'll see any warnings that are emitted during the build process. Without it, relevant warnings often don't appear. Cheers, Rob | [reply] [d/l] [select] |
|
Re^6: Inline::C produces compilation errors
by holandes777 (Scribe) on Nov 24, 2007 at 13:07 UTC | |
output of perl -V is :
| [reply] [d/l] |
by syphilis (Archbishop) on Nov 24, 2007 at 13:59 UTC | |
What error messages do you get when you try to build a module (other than Inline) that includes XS code. (It's generally easier to diagnose error messages in relation to XS modules - rather than trying to diagnose the errors that Inline spits out.) Mind you, I'm pretty much at a loss to see why you should be having such difficulty. The output of your 'perl -V' is very similar to mine for linux. My configure args were '-des' and the '-Wdeclaration-after-statement' is missing on my build. Apart from that I can't spot a significant difference. I don't think the '-des' vs '-de' difference could amount to much ... I have no idea what the '-Wdeclaration-after-statement' stuff is about (and whether it could be relevant). Anyway, if you like to post the results of trying to build an XS module (such as Math::FFT), then I'll certainly take a look. (Whether I'll be able to provide some useful assistance is another question .... looking more'n'more unlikely :-) Cheers, Rob | [reply] |