Noissue - Use clang++ and libc++ on Linux

Noissue - Use clang++ and libc++ on Linux

[sergei, 2018-06-28 11:54:16]

looks good, though could you please also update readme regarding libc++ and
confirm that travis-ci runs it successfully.

Additionally, I just wonder whether it's possible to compile V8 not with libc++
which comes with V8 but with another lib. It can be a hurdle for those who just
wants to add libadblockplus to a native application already based on libstdc++.

@Anton, merely FYI, it seems now we require to use libc++ everywhere (it is
already used by default in android ndk).

[anton, 2018-06-28 12:02:23]

On 2018/06/28 11:54:16, sergei wrote:
> looks good, though could you please also update readme regarding libc++ and
> confirm that travis-ci runs it successfully.
> 
> Additionally, I just wonder whether it's possible to compile V8 not with
libc++
> which comes with V8 but with another lib. It can be a hurdle for those who
just
> wants to add libadblockplus to a native application already based on
libstdc++.
> 
> @Anton, merely FYI, it seems now we require to use libc++ everywhere (it is
> already used by default in android ndk).

does /can it conflict with chromium in any way?

[sergei, 2018-06-28 12:08:50]

On 2018/06/28 12:02:23, anton wrote:
> On 2018/06/28 11:54:16, sergei wrote:
> > looks good, though could you please also update readme regarding libc++ and
> > confirm that travis-ci runs it successfully.
> > 
> > Additionally, I just wonder whether it's possible to compile V8 not with
> libc++
> > which comes with V8 but with another lib. It can be a hurdle for those who
> just
> > wants to add libadblockplus to a native application already based on
> libstdc++.
> > 
> > @Anton, merely FYI, it seems now we require to use libc++ everywhere (it is
> > already used by default in android ndk).
> 
> does /can it conflict with chromium in any way?

since chromium uses it too, it should not conflict. BTW, I start to think more
about using of NDK from chromium when we compile libadblockplus* for chromium
since the downloaded NDK sometimes differs from the one which comes with the
chromium. Perhaps we could integrate it in CI of libadblockplus and
libadblockplus-android, we just need to check how to obtain NDK from chromium
without chromium.
Additionally, it means that what we test on linux now is closer to what works on
android because any libc++ is closer to some another libc++ than to any
libstdc++.

[anton, 2018-06-28 12:12:31]

On 2018/06/28 12:08:50, sergei wrote:
we just need to check how to obtain NDK from chromium
> without chromium.
it's easy:
https://gitlab.com/eyeo/adblockplus/chromium/blob/issue-9/DEPS#L617

[anton, 2018-06-28 12:14:24]

On 2018/06/28 12:12:31, anton wrote:
> On 2018/06/28 12:08:50, sergei wrote:
> we just need to check how to obtain NDK from chromium
> > without chromium.
> it's easy:
> https://gitlab.com/eyeo/adblockplus/chromium/blob/issue-9/DEPS#L617

better to use think branch:
https://gitlab.com/eyeo/adblockplus/chromium/blob/dev-67.0.3396.87/DEPS#L613

[hub, 2018-06-28 12:18:08]

On 2018/06/28 11:54:16, sergei wrote:
> looks good, though could you please also update readme regarding libc++ and
> confirm that travis-ci runs it successfully.

I already did. See
https://travis-ci.org/hfiguiere/libadblockplus/builds/397592090
(only difference is that I call make V=1)

> Additionally, I just wonder whether it's possible to compile V8 not with
libc++
> which comes with V8 but with another lib. It can be a hurdle for those who
just
> wants to add libadblockplus to a native application already based on
libstdc++.

We can evaluate that as a second step.

[sergei, 2018-06-28 12:21:34]

On 2018/06/28 12:18:08, hub wrote:
> On 2018/06/28 11:54:16, sergei wrote:
> > looks good, though could you please also update readme regarding libc++ and
> > confirm that travis-ci runs it successfully.
Cool.
> 
> I already did. See
> https://travis-ci.org/hfiguiere/libadblockplus/builds/397592090
> (only difference is that I call make V=1)
> 
> > Additionally, I just wonder whether it's possible to compile V8 not with
> libc++
> > which comes with V8 but with another lib. It can be a hurdle for those who
> just
> > wants to add libadblockplus to a native application already based on
> libstdc++.
> 
> We can evaluate that as a second step.
Sure.

I think we should just say in readme, that we normally work with libc++ because
V8 usually requires it. If someone wants to use something else, e.g. libstdc++,
then we didn't test it but it should work. In case of any issues send a ticket
as with other compilers.

[hub, 2018-06-28 14:09:03]

updated patch

[sergei, 2018-06-28 15:10:44]

LGTM

[anton, 2018-07-02 10:12:38]

https://codereview.adblockplus.org/29817566/diff/29818560/README.md
File README.md (right):

https://codereview.adblockplus.org/29817566/diff/29818560/README.md#newcode33
README.md:33: * clang 5.5
in travis config it's clang 4.0 required, but in README it's 5.5. Which one?
Also in travis g++ is still required (`g++-7`), but here it's just deleted from
requirements.

[hub, 2018-07-04 07:50:50]

updated patch

https://codereview.adblockplus.org/29817566/diff/29818560/README.md
File README.md (right):

https://codereview.adblockplus.org/29817566/diff/29818560/README.md#newcode33
README.md:33: * clang 5.5
On 2018/07/02 10:12:38, anton wrote:
> in travis config it's clang 4.0 required, but in README it's 5.5. Which one?
> Also in travis g++ is still required (`g++-7`), but here it's just deleted
from
> requirements.

I have changed it to 5.0. And remove the explicit dependencies for clang and gcc
in the travis.yml. 5.0 is what is provided if you specify clang as the C++
compiler.

I have tested it on TravisCI.

[sergei, 2018-07-04 07:53:16]

LGTM

[anton, 2018-07-04 08:00:40]

On 2018/07/04 07:50:50, hub wrote:
> updated patch
> 
> https://codereview.adblockplus.org/29817566/diff/29818560/README.md
> File README.md (right):
> 
> https://codereview.adblockplus.org/29817566/diff/29818560/README.md#newcode33
> README.md:33: * clang 5.5
> On 2018/07/02 10:12:38, anton wrote:
> > in travis config it's clang 4.0 required, but in README it's 5.5. Which one?
> > Also in travis g++ is still required (`g++-7`), but here it's just deleted
> from
> > requirements.
> 
> I have changed it to 5.0. And remove the explicit dependencies for clang and
gcc
> in the travis.yml. 5.0 is what is provided if you specify clang as the C++
> compiler.
> 
> I have tested it on TravisCI.

LGTM though having not too deep understanding