> -----Original Message-----
> From: Khem Raj <[email protected]>
> Sent: den 16 april 2020 17:49
> To: Trevor Gamblin <[email protected]>
> Cc: Peter Kjellerstedt <[email protected]>; openembeded-devel
> <[email protected]>
> Subject: Re: [oe] [meta-webserver][PATCH] apache2: upgrade 2.4.41 ->
> 2.4.43
>
> On Thu, Apr 16, 2020 at 6:21 AM Trevor Gamblin
> <[email protected]> wrote:
> >
> > On 4/15/20 7:02 PM, Peter Kjellerstedt wrote:
> > >> -----Original Message-----
> > >> From: [email protected] <openembedded-
> > >> [email protected]> On Behalf Of Trevor Gamblin
> > >> Sent: den 15 april 2020 02:10
> > >> To: Khem Raj <[email protected]>
> > >> Cc: openembeded-devel <[email protected]>
> > >> Subject: Re: [oe] [meta-webserver][PATCH] apache2: upgrade 2.4.41 ->
> > >> 2.4.43
> > >>
> > >> On 4/14/20 6:38 PM, Khem Raj wrote:
> > >>> apache-websockets does not like it
> > >>>
> > >>> https://errors.yoctoproject.org/Errors/Details/403771/
> > >>
> > >> Thanks, will take a look at it and re-send.
> > >
> > > It appears they have added cross compilation support to apxs, but
> > > it clashes with the changes done in OE to make it support cross
> > > compilation. I tried to solve it by basically removing the OE
> > > changes, but it was not as simple as that. In the end we worked
> > > around it with this quick fix:
> > >
> > > apache_sysroot_preprocess_append() {
> > > # Disable apxs' built in cross compilation support and rely on
> > > what the
> > > # recipe sets up.
> > > sed -i 's!$destdir = .*!$destdir = "";!'
> > > ${SYSROOT_DESTDIR}${bindir_crossscripts}/apxs
> > > }
> > >
> > > However, I hope you can come up with a better solution that
> > > actually makes use of what they did upstream and thus lessen
> > > the changes that have to be applied by OE.
> >
> > Hi Peter,
> >
> > Ironically I'm seeing your email just as I submit a v2 reverting their
> > cross-compilation changes. I wasn't able to find a better workaround
> > than just reverting the destdir stuff that they added, but I did it via
> > a patch file instead of within the recipe itself. I do like your
> > approach better though, as it avoids adding a patch file that'll have
> > to be maintained. Do you want to submit your fix in lieu of mine?
>
> yeah I think revert is last resort. So lets see if we can fix it
Well, my solution just disables the upstream change, so it doesn't
really make use of it. I think a patch is better than the quick sed
fix I did since it means there is context to the change. However,
you don't need to revert all of the upstream change. Just commenting
out these lines would do:
my $ddi = rindex($0, "@exp_bindir@");
if ($ddi >= 0) {
$destdir = substr($0, 0, $ddi);
}
That should improve the chance of the patch continuing to apply for
future versions of apache.
//Peter
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#83976):
https://lists.openembedded.org/g/openembedded-devel/message/83976
Mute This Topic: https://lists.openembedded.org/mt/73018511/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-devel/unsub
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
