[oe] pcmciautils: fix build to use flex from sysroots, not from host
dbaryshkov at gmail.com
Mon Jun 7 00:20:21 CEST 2010
On Sun, Jun 6, 2010 at 10:37 PM, Khem Raj <raj.khem at gmail.com> wrote:
> On Sun, Jun 6, 2010 at 4:00 AM, Dmitry Eremin-Solenikov
> <dbaryshkov at gmail.com> wrote:
>> On Sat, Jun 5, 2010 at 3:24 AM, Khem Raj <raj.khem at gmail.com> wrote:
>>> This commit
>>> commit c65ee13298e8c22274b5bcabe06f2116b8b7ebf2
>>> Author: Dmitry Eremin-Solenikov <dbaryshkov at gmail.com>
>>> AuthorDate: Fri Jun 4 16:26:17 2010 +0400
>>> Commit: Dmitry Eremin-Solenikov <dbaryshkov at gmail.com>
>>> CommitDate: Fri Jun 4 16:44:18 2010 +0400
>>> pcmciautils: fix build to use flex from sysroots, not from host
>>> Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov at gmail.com>
>>> breaks the build when I have flex-native in ASSUME_PROVIDED.
>>> It there any specific reason why one should use flex from hosted sysroot ?
>>> is the flex package on all distribution broken for some reason ?
>> Then we either:
>> 1) have to put flex in prerequisites, check it in sanity.bbclass (and
>> drop flex-native)
>> 2) have to fix pcmciautils to detect path to flex
>> 3) expand some class to autodetect the flex/bison/etc. programs
>> Because currently pcmciautils just failed for me because I was building them on
>> the box where I had no flex installed, but build (via some other dependancy)
>> in OE sysroot.
> Well then you only need it in DEPENDS of pcmciautils thats all and not
> hardcode it
Fine with me. I'm commiting a patch which will drop the hardcoded path
and use just
LEX=flex. That should be fine for both our cases, shan't it?
>> With best wishes
With best wishes
More information about the Openembedded-devel