Re: [Spice-devel] [PATCH spice-common] Add .gitlab-ci.yml file

2018-06-06 Thread Frediano Ziglio 
> 
> Signed-off-by: Eduardo Lima (Etrunko) 
> ---
>  .gitlab-ci.yml | 18 ++
>  1 file changed, 18 insertions(+)
>  create mode 100644 .gitlab-ci.yml
> 
> diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
> new file mode 100644
> index 000..1d61805
> --- /dev/null
> +++ b/.gitlab-ci.yml
> @@ -0,0 +1,18 @@
> +image: fedora:latest
> +
> +before_script:
> +  - >
> +dnf install 'dnf-command(copr)' git libtool make libasan
> +python3 python3-six python3-pyparsing glib-networking
> +-y
> +  - dnf copr enable @spice/nightly -y
> +  - dnf builddep spice -y
> +
> +makecheck:
> +  script:
> +  - >
> +CFLAGS='-O2 -pipe -g -fsanitize=address -fno-omit-frame-pointer
> -Wframe-larger-than=40920'
> +LDFLAGS='-fsanitize=address -lasan'
> +./autogen.sh --enable-extra-checks
> +  - make
> +  - make check || (cat tests/test-suite.log && exit 1)

Acked, no much discussions beside other build comments.

Frediano
___
Spice-devel mailing list
Spice-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/spice-devel

Re: [Spice-devel] [PATCH spice-common] Add .gitlab-ci.yml file

2018-05-18 Thread Eduardo Lima (Etrunko) 
On 18/05/18 05:57, Frediano Ziglio wrote:
>>
>> On 17/05/18 12:11, Frediano Ziglio wrote:

 Here is an example of a successful run:

 https://gitlab.freedesktop.org/etrunko/spice-common/-/jobs/549

>>>
>>> Stupid question. Is it worth doing giving that spice-common is compiled
>>> and tested from both spice-server and spice-gtk?
>>>
>>
>> Not a stupid question at all, the main reason for this one is actually
>> because meson introduced partial builds of the spice-common code base.
>> If we are building server, only server code is generated, while if
>> building spice-gtk only client code is generated.
>>
> 
> My we need all these libraries and separation? Won't be easier to have
> a single library and let the linker do his job?
> Protocol file should not be changed that often so code generation and
> compile should not be that frequent.

I just followed the autotools build, but that can be changed for sure.
Currently we build 3 libraries, one common, server and client (both
linking to common).

> I think would be faster if the python script could generate multiple
> files instead of having the parsing done for every file.
> 

That would help yes.


>> Now, with the introduction of the test-overflow, which depends on both
>> server and client marshallers, it is not possible to build this tests
>> with neither server or gtk, when using meson.
>>
> 
> Ok, the problem is the way currently meson build script works.
> 

Yes, I added the option to build either server or client code depending
on which repository you are working on, avoiding generating code,
compling and linking it. It can save some precious seconds every build.

>> Besides that, I think it will be useful, since we usually push lots of
>> patches to spice-common before updating the submodules in spice server
>> and spice-gtk. This would enable us to find errors in tests before that
>> update happens, although we could definitely use some more tests in
>> spice-common. ;)
>>
> 
> That make sense!
> 
>> Anyway, if you think it is too much hassle, I can force building both
>> server and client marshallers with meson for both spice-server and
>> spice-gtk.
>>
> 
> I personally would prefer to have a single library.
> 

That is not what happens at the moment, even if we generate and build
code for both server and client, like I explained above. So I assume you
are talking about having spice common build both server and client code
independently of which project you are working on, right?

If so, I will keep that option there so people can change it locally but
will always pass -Dgenerate_code="all" when building both server and client.


 On 17/05/18 11:29, Eduardo Lima (Etrunko) wrote:
> Signed-off-by: Eduardo Lima (Etrunko) 
> ---
>  .gitlab-ci.yml | 18 ++
>  1 file changed, 18 insertions(+)
>  create mode 100644 .gitlab-ci.yml
>
> diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
> new file mode 100644
> index 000..1d61805
> --- /dev/null
> +++ b/.gitlab-ci.yml
> @@ -0,0 +1,18 @@
> +image: fedora:latest
> +
> +before_script:
> +  - >
> +dnf install 'dnf-command(copr)' git libtool make libasan
> +python3 python3-six python3-pyparsing glib-networking
> +-y
> +  - dnf copr enable @spice/nightly -y
> +  - dnf builddep spice -y
> +
> +makecheck:
> +  script:
> +  - >
> +CFLAGS='-O2 -pipe -g -fsanitize=address -fno-omit-frame-pointer
> -Wframe-larger-than=40920'
> +LDFLAGS='-fsanitize=address -lasan'
> +./autogen.sh --enable-extra-checks
> +  - make
> +  - make check || (cat tests/test-suite.log && exit 1)
>

>>>
>>> Frediano
>>>
>>
>>
>> --
>> Eduardo de Barros Lima (Etrunko)
>> Software Engineer - RedHat
>> etru...@redhat.com
>>


-- 
Eduardo de Barros Lima (Etrunko)
Software Engineer - RedHat
etru...@redhat.com
___
Spice-devel mailing list
Spice-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/spice-devel

Re: [Spice-devel] [PATCH spice-common] Add .gitlab-ci.yml file

2018-05-18 Thread Frediano Ziglio 
> 
> On 17/05/18 12:11, Frediano Ziglio wrote:
> >>
> >> Here is an example of a successful run:
> >>
> >> https://gitlab.freedesktop.org/etrunko/spice-common/-/jobs/549
> >>
> > 
> > Stupid question. Is it worth doing giving that spice-common is compiled
> > and tested from both spice-server and spice-gtk?
> > 
> 
> Not a stupid question at all, the main reason for this one is actually
> because meson introduced partial builds of the spice-common code base.
> If we are building server, only server code is generated, while if
> building spice-gtk only client code is generated.
> 

My we need all these libraries and separation? Won't be easier to have
a single library and let the linker do his job?
Protocol file should not be changed that often so code generation and
compile should not be that frequent.
I think would be faster if the python script could generate multiple
files instead of having the parsing done for every file.

> Now, with the introduction of the test-overflow, which depends on both
> server and client marshallers, it is not possible to build this tests
> with neither server or gtk, when using meson.
> 

Ok, the problem is the way currently meson build script works.

> Besides that, I think it will be useful, since we usually push lots of
> patches to spice-common before updating the submodules in spice server
> and spice-gtk. This would enable us to find errors in tests before that
> update happens, although we could definitely use some more tests in
> spice-common. ;)
> 

That make sense!

> Anyway, if you think it is too much hassle, I can force building both
> server and client marshallers with meson for both spice-server and
> spice-gtk.
> 

I personally would prefer to have a single library.

> >> On 17/05/18 11:29, Eduardo Lima (Etrunko) wrote:
> >>> Signed-off-by: Eduardo Lima (Etrunko) 
> >>> ---
> >>>  .gitlab-ci.yml | 18 ++
> >>>  1 file changed, 18 insertions(+)
> >>>  create mode 100644 .gitlab-ci.yml
> >>>
> >>> diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
> >>> new file mode 100644
> >>> index 000..1d61805
> >>> --- /dev/null
> >>> +++ b/.gitlab-ci.yml
> >>> @@ -0,0 +1,18 @@
> >>> +image: fedora:latest
> >>> +
> >>> +before_script:
> >>> +  - >
> >>> +dnf install 'dnf-command(copr)' git libtool make libasan
> >>> +python3 python3-six python3-pyparsing glib-networking
> >>> +-y
> >>> +  - dnf copr enable @spice/nightly -y
> >>> +  - dnf builddep spice -y
> >>> +
> >>> +makecheck:
> >>> +  script:
> >>> +  - >
> >>> +CFLAGS='-O2 -pipe -g -fsanitize=address -fno-omit-frame-pointer
> >>> -Wframe-larger-than=40920'
> >>> +LDFLAGS='-fsanitize=address -lasan'
> >>> +./autogen.sh --enable-extra-checks
> >>> +  - make
> >>> +  - make check || (cat tests/test-suite.log && exit 1)
> >>>
> >>
> > 
> > Frediano
> > 
> 
> 
> --
> Eduardo de Barros Lima (Etrunko)
> Software Engineer - RedHat
> etru...@redhat.com
> 
___
Spice-devel mailing list
Spice-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/spice-devel

Re: [Spice-devel] [PATCH spice-common] Add .gitlab-ci.yml file

2018-05-17 Thread Eduardo Lima (Etrunko) 
On 17/05/18 12:11, Frediano Ziglio wrote:
>>
>> Here is an example of a successful run:
>>
>> https://gitlab.freedesktop.org/etrunko/spice-common/-/jobs/549
>>
> 
> Stupid question. Is it worth doing giving that spice-common is compiled
> and tested from both spice-server and spice-gtk?
> 

Not a stupid question at all, the main reason for this one is actually
because meson introduced partial builds of the spice-common code base.
If we are building server, only server code is generated, while if
building spice-gtk only client code is generated.

Now, with the introduction of the test-overflow, which depends on both
server and client marshallers, it is not possible to build this tests
with neither server or gtk, when using meson.

Besides that, I think it will be useful, since we usually push lots of
patches to spice-common before updating the submodules in spice server
and spice-gtk. This would enable us to find errors in tests before that
update happens, although we could definitely use some more tests in
spice-common. ;)

Anyway, if you think it is too much hassle, I can force building both
server and client marshallers with meson for both spice-server and
spice-gtk.

>> On 17/05/18 11:29, Eduardo Lima (Etrunko) wrote:
>>> Signed-off-by: Eduardo Lima (Etrunko) 
>>> ---
>>>  .gitlab-ci.yml | 18 ++
>>>  1 file changed, 18 insertions(+)
>>>  create mode 100644 .gitlab-ci.yml
>>>
>>> diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
>>> new file mode 100644
>>> index 000..1d61805
>>> --- /dev/null
>>> +++ b/.gitlab-ci.yml
>>> @@ -0,0 +1,18 @@
>>> +image: fedora:latest
>>> +
>>> +before_script:
>>> +  - >
>>> +dnf install 'dnf-command(copr)' git libtool make libasan
>>> +python3 python3-six python3-pyparsing glib-networking
>>> +-y
>>> +  - dnf copr enable @spice/nightly -y
>>> +  - dnf builddep spice -y
>>> +
>>> +makecheck:
>>> +  script:
>>> +  - >
>>> +CFLAGS='-O2 -pipe -g -fsanitize=address -fno-omit-frame-pointer
>>> -Wframe-larger-than=40920'
>>> +LDFLAGS='-fsanitize=address -lasan'
>>> +./autogen.sh --enable-extra-checks
>>> +  - make
>>> +  - make check || (cat tests/test-suite.log && exit 1)
>>>
>>
> 
> Frediano
> 


-- 
Eduardo de Barros Lima (Etrunko)
Software Engineer - RedHat
etru...@redhat.com
___
Spice-devel mailing list
Spice-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/spice-devel

Re: [Spice-devel] [PATCH spice-common] Add .gitlab-ci.yml file

2018-05-17 Thread Frediano Ziglio 
> 
> Here is an example of a successful run:
> 
> https://gitlab.freedesktop.org/etrunko/spice-common/-/jobs/549
> 

Stupid question. Is it worth doing giving that spice-common is compiled
and tested from both spice-server and spice-gtk?

> On 17/05/18 11:29, Eduardo Lima (Etrunko) wrote:
> > Signed-off-by: Eduardo Lima (Etrunko) 
> > ---
> >  .gitlab-ci.yml | 18 ++
> >  1 file changed, 18 insertions(+)
> >  create mode 100644 .gitlab-ci.yml
> > 
> > diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
> > new file mode 100644
> > index 000..1d61805
> > --- /dev/null
> > +++ b/.gitlab-ci.yml
> > @@ -0,0 +1,18 @@
> > +image: fedora:latest
> > +
> > +before_script:
> > +  - >
> > +dnf install 'dnf-command(copr)' git libtool make libasan
> > +python3 python3-six python3-pyparsing glib-networking
> > +-y
> > +  - dnf copr enable @spice/nightly -y
> > +  - dnf builddep spice -y
> > +
> > +makecheck:
> > +  script:
> > +  - >
> > +CFLAGS='-O2 -pipe -g -fsanitize=address -fno-omit-frame-pointer
> > -Wframe-larger-than=40920'
> > +LDFLAGS='-fsanitize=address -lasan'
> > +./autogen.sh --enable-extra-checks
> > +  - make
> > +  - make check || (cat tests/test-suite.log && exit 1)
> > 
> 

Frediano
___
Spice-devel mailing list
Spice-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/spice-devel

Re: [Spice-devel] [PATCH spice-common] Add .gitlab-ci.yml file

2018-05-17 Thread Eduardo Lima (Etrunko) 
Here is an example of a successful run:

https://gitlab.freedesktop.org/etrunko/spice-common/-/jobs/549

On 17/05/18 11:29, Eduardo Lima (Etrunko) wrote:
> Signed-off-by: Eduardo Lima (Etrunko) 
> ---
>  .gitlab-ci.yml | 18 ++
>  1 file changed, 18 insertions(+)
>  create mode 100644 .gitlab-ci.yml
> 
> diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
> new file mode 100644
> index 000..1d61805
> --- /dev/null
> +++ b/.gitlab-ci.yml
> @@ -0,0 +1,18 @@
> +image: fedora:latest
> +
> +before_script:
> +  - >
> +dnf install 'dnf-command(copr)' git libtool make libasan
> +python3 python3-six python3-pyparsing glib-networking
> +-y
> +  - dnf copr enable @spice/nightly -y
> +  - dnf builddep spice -y
> +
> +makecheck:
> +  script:
> +  - >
> +CFLAGS='-O2 -pipe -g -fsanitize=address -fno-omit-frame-pointer 
> -Wframe-larger-than=40920'
> +LDFLAGS='-fsanitize=address -lasan'
> +./autogen.sh --enable-extra-checks
> +  - make
> +  - make check || (cat tests/test-suite.log && exit 1)
> 


-- 
Eduardo de Barros Lima (Etrunko)
Software Engineer - RedHat
etru...@redhat.com
___
Spice-devel mailing list
Spice-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/spice-devel