[gentoo-devhelp] ebuild environment

[gentoo-devhelp] ebuild environment

[Thomas Kahle]

[-- Attachment #1: Type: text/plain, Size: 935 bytes --]

Hi,

I'm having some autotools trouble. In an ebuild I use sed to modify
configure.ac with something like

sed "s/\"-lntl\"/\"-lntl\" \"-lgf2x -lntl\"/" -i "${S}/configure.ac"

make configure is run by the ebuild after this! If I run configure
outside the portage environment this test succeeds with "-lgf2x -lntl"
and everything is fine. However, if run from portage the test fails. Now
I wonder if I have to do something special. I even used grep and emerge
--debug to confirm that the configure script actually has the modified
test. It just seems that the library -lgf2x can not be found inside the
portage environment.

Any ideas?
Thanks
Thomas



--
Thomas Kahle

The fundamental theorem of algebra is open source. Like any other
mathematical theorem it can be applied free of charge and everybody
has access to its proof and can convince himself how it works. Why
should software be any different?


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 261 bytes --]

[gentoo-devhelp] Re: ebuild environment

[Nikos Chantziaras]

On 08/11/2009 11:42 PM, Thomas Kahle wrote:
> Hi,
>
> I'm having some autotools trouble. In an ebuild I use sed to modify
> configure.ac with something like
>
> sed "s/\"-lntl\"/\"-lntl\" \"-lgf2x -lntl\"/" -i "${S}/configure.ac"
>
> make configure is run by the ebuild after this!

"make configure"?  Don't you need eautoreconf?

Re: [gentoo-devhelp] Re: ebuild environment

[Thomas Kahle]

[-- Attachment #1: Type: text/plain, Size: 1144 bytes --]

Nikos Chantziaras wrote:
> On 08/11/2009 11:42 PM, Thomas Kahle wrote:
>> Hi,
>>
>> I'm having some autotools trouble. In an ebuild I use sed to modify
>> configure.ac with something like
>>
>> sed "s/\"-lntl\"/\"-lntl\" \"-lgf2x -lntl\"/" -i "${S}/configure.ac"
>>
>> make configure is run by the ebuild after this!
> 
> "make configure"?  Don't you need eautoreconf?
I dont know. I suppose not. How this package is supposed to be compiled is:

make configure
configure
make
make install

since I do the sed'ing of configure.ac before 'make configure' (which
created 'configure' script in the first place) I assume that eautoreconf
is not needed. I still tried it in 'various places' but no success. Let
me say again that I checked that the configure script is actually
modified and running the parts of the ebuild outside of portage works as
it should.

-- 
Thomas Kahle

The fundamental theorem of algebra is open source. Like any other
mathematical theorem it can be applied free of charge and everybody
has access to its proof and can convince himself how it works. Why
should software be any different?


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 261 bytes --]

Re: [gentoo-devhelp] Re: ebuild environment

[Thomas Kahle]

[-- Attachment #1: Type: text/plain, Size: 1286 bytes --]

Thomas Kahle wrote:
> Nikos Chantziaras wrote:
>> On 08/11/2009 11:42 PM, Thomas Kahle wrote:
>>> Hi,
>>>
>>> I'm having some autotools trouble. 

Ok, got it. It is an "--as-needed" thing. Now god, help me with this...


In an ebuild I use sed to modify
>>> configure.ac with something like
>>>
>>> sed "s/\"-lntl\"/\"-lntl\" \"-lgf2x -lntl\"/" -i "${S}/configure.ac"
>>>
>>> make configure is run by the ebuild after this!
>> "make configure"?  Don't you need eautoreconf?
> I dont know. I suppose not. How this package is supposed to be compiled is:
> 
> make configure
> configure
> make
> make install
> 
> since I do the sed'ing of configure.ac before 'make configure' (which
> created 'configure' script in the first place) I assume that eautoreconf
> is not needed. I still tried it in 'various places' but no success. Let
> me say again that I checked that the configure script is actually
> modified and running the parts of the ebuild outside of portage works as
> it should.
> 


-- 
Thomas Kahle

The fundamental theorem of algebra is open source. Like any other
mathematical theorem it can be applied free of charge and everybody
has access to its proof and can convince himself how it works. Why
should software be any different?


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 261 bytes --]