Re: Qt and SecureTransport vs. OpenSSL

2017-01-21 Thread René J . V . Bertin 
On Saturday January 21 2017 15:21:19 Marcus Calhoun-Lopez wrote:


> My current plan is to use Secure Transport instead and create an openssl 
> variant.

Until we know more about the pros and cons it may be preferable to make a 
securetransport variant, or else make the openssl variant the default.

Qt also has a dynamic openssl mode, in which it loads libssl at runtime. I 
reckon that it should be possible to ensure it uses the one from MacPorts. I'll 
try to look into that one of these days.

R.

Re: Qt and SecureTransport vs. OpenSSL

2017-01-21 Thread Marcus Calhoun-Lopez 
The openssl dependency is one reason that qt5-qtcreator does not build on the 
buildbots.
My current plan is to use Secure Transport instead and create an openssl 
variant.
Unfortunately, my knowledge of Transport Layer Security is limited, so I do not 
know if there are other considerations.

-Marcus

> On Jan 21, 2017, at 11:56 AM, René J.V. Bertin  wrote:
> 
> Hi,
> 
> Somehow I never noticed this before:
> 
> - Qt4 and Qt5 are configured with -openssl-linked to ensure the MacPorts 
> version is used
> - this disables SecureTransport support which Qt normally uses on Mac
> 
> I guess that using OpenSSL means we get whatever more recent SSL is in 
> MacPorts than the host's SecureTransport layer, and that that's an advantage?
> 
> But isn't there a licensing/redistribution implication for binaries that link 
> to OpenSSL?
> 
> R.

Qt and SecureTransport vs. OpenSSL

2017-01-21 Thread René J . V . Bertin 
Hi,

Somehow I never noticed this before:

- Qt4 and Qt5 are configured with -openssl-linked to ensure the MacPorts 
version is used
- this disables SecureTransport support which Qt normally uses on Mac

I guess that using OpenSSL means we get whatever more recent SSL is in MacPorts 
than the host's SecureTransport layer, and that that's an advantage?

But isn't there a licensing/redistribution implication for binaries that link 
to OpenSSL?

R.

Re: [macports-ports] branch master updated: fontconfig: support mpkg/mdmg packaging, add postinstall script to initialize/update font cache. Fixes #53350

2017-01-21 Thread Rainer Müller 
On 2017-01-21 15:42, Kevin Reid wrote:
> On Sat, Jan 21, 2017 at 5:20 AM, Ryan Schmidt  > wrote:
> 
> GitHub doesn't have server-side pre-commit hooks.
> 
> 
> It does have pre-(merge or push) hooks, in the form of “status checks”.

As far as I understand it, GitHub offers that only for pull requests,
which will not check any commits that are pushed directly. Is there
anything we have been missing?

Direct pushes are the normal way we are working, because submitted pull
requests usually need additional work before they can be accepted into
master (like squashing commits, fixing commit messages, etc.).

A pre-commit hook in the local git repostiory requires manual setup,
which will probably only be done by those who are already careful.

Personally, I would not want to make GitHub API calls every time I make
a local git commit. Especially not when I am doing intermediate commits
that I am later squashing before pushing.

Rainer

Re: [macports-ports] branch master updated: fontconfig: support mpkg/mdmg packaging, add postinstall script to initialize/update font cache. Fixes #53350

2017-01-21 Thread Kevin Reid 
On Sat, Jan 21, 2017 at 5:20 AM, Ryan Schmidt 
wrote:

> GitHub doesn't have server-side pre-commit hooks.
>

It does have pre-(merge or push) hooks, in the form of “status checks”.

Re: port:mpv issues

2017-01-21 Thread Marko Käning 
Hi Mihai,

On 21 Jan 2017, at 14:15 , Mihai Moldovan  wrote:
> Done now.

Thanks.


> Please always include the OS X/macOS version you're running the program under.
> You can ignore the deprecation messages, these aren't changing mpv's 
> behaviour.

Ah… Just realising that this could be due to me trying to run MPV on an "El 
Capitan" VM…


> I know that error message from my 10.6 VM (and maybe also 10.7, though I have
> never tested on that platform) - I don't know why it doesn't like the pixel
> format there, though.
> 
> In general, this seems to be an OpenGL issue, or rather the error message that
> the OS returns when trying to create the OpenGL context. I have so far been
> unable to find out, why that is, though.

Yeah, had that already with VLC in the past on such VMs…


> If you do see this failure on 10.8 or higher, I have absolutely no clue what
> could be wrong. Is 3D hardware acceleration in general working?

In VMs the 3D accel might not be fully supported. (Am here on Parallels 11.)

Sorry for the noise regarding this.

Marko

Re: [macports-ports] branch master updated: fontconfig: support mpkg/mdmg packaging, add postinstall script to initialize/update font cache. Fixes #53350

2017-01-21 Thread Marko Käning 

On 21 Jan 2017, at 14:20 , Ryan Schmidt  wrote:

> GitHub doesn't have server-side pre-commit hooks. 

We need them client-side, of course.

Re: [macports-ports] branch master updated: fontconfig: support mpkg/mdmg packaging, add postinstall script to initialize/update font cache. Fixes #53350

2017-01-21 Thread Ryan Schmidt 


> On Jan 21, 2017, at 06:46, Marko Käning  wrote:
> 
> Hi Ryan,
> 
>> On 20 Jan 2017, at 05:13 , Ryan Schmidt  wrote:
>> 
>> Remember that now that we've switched to GitHub, "#53350" refers to *GitHub 
>> pull request* 53350.
> 
> this is why I suggested a pre-commit sanity check of commit messages in 
> 
>https://trac.macports.org/ticket/52763
> 
> I admit that it might not be so easily implementable, but still, for commits 
> like this it seems to make a lot of sense, since you don’t want to mess up a 
> projects’ commit history by filling it with bogus links to “PRs” which are in 
> fact original MacPorts trac ticket references.
> 
> Greets,
> Marko

GitHub doesn't have server-side pre-commit hooks.

Re: port:mpv issues

2017-01-21 Thread Mihai Moldovan 
On 21.01.2017 01:59 PM, Marko Käning wrote:
> In there /opt/local/share/examples/mpv/config-maintainer is referred to,
> but in fact that file isn’t to be found there, but at
> 
>   /opt/local/share/doc/mpv/examples/config-maintainer
> 
> i.e. you’re missing the "/doc/mpv” bit in there…

I forgot to update the path in the notes message when I fixed the port to comply
with https://trac.macports.org/ticket/47690

Done now.


> Apart from that I ran into this:
> ---
> $ mpv foo.mp4 
> --vo=opengl-hq is deprecated! Use --profile=opengl-hq instead.
> Playing: foo.mp4
>  (+) Video --vid=1 (*) (h264)
>  (+) Audio --aid=1 --alang=und (*) (aac)
> [vo/opengl] Using suboptions is deprecated. Use the global '--opengl-backend' 
> option instead of 'backend' suboption.
> [vo/opengl] error creating CGL context: invalid pixel format (10002)
> Error opening/initializing the selected video_out (-vo) device.
> Video: no video
> 
> 
> Exiting... (Errors when loading file)
> ---
> 
> Any ideas what all that means?

Please always include the OS X/macOS version you're running the program under.
You can ignore the deprecation messages, these aren't changing mpv's behavior.

I know that error message from my 10.6 VM (and maybe also 10.7, though I have
never tested on that platform) - I don't know why it doesn't like the pixel
format there, though.

In general, this seems to be an OpenGL issue, or rather the error message that
the OS returns when trying to create the OpenGL context. I have so far been
unable to find out, why that is, though.

If you do see this failure on 10.8 or higher, I have absolutely no clue what
could be wrong. Is 3D hardware acceleration in general working?



Mihai





signature.asc
Description: OpenPGP digital signature

port:mpv issues

2017-01-21 Thread Marko Käning 
Hi Mihai,

I noticed these messages during installation of port mpv:

--->  Activating mpv 
@0.22.0_2+audiocd+bundle+dvd+libarchive+network+opengl+osd+python27+rubberband

The maintainer ships his personal config file in
* /opt/local/share/examples/mpv/config-maintainer.

You are strongly encouraged to copy this file to either
* /opt/local/etc/mpv/mpv.conf (globally for all users of the system) or
* ~/.mpv/config (only for your current user)
and adjust it to your needs.
This config file already defines the necessary video output settings to make 
use of the hardware acceleration features provided by your operating system.

--->  Cleaning mpv

In there /opt/local/share/examples/mpv/config-maintainer is referred to,
but in fact that file isn’t to be found there, but at

/opt/local/share/doc/mpv/examples/config-maintainer

i.e. you’re missing the "/doc/mpv” bit in there…


...

Apart from that I ran into this:
---
$ mpv foo.mp4 
--vo=opengl-hq is deprecated! Use --profile=opengl-hq instead.
Playing: foo.mp4
 (+) Video --vid=1 (*) (h264)
 (+) Audio --aid=1 --alang=und (*) (aac)
[vo/opengl] Using suboptions is deprecated. Use the global '--opengl-backend' 
option instead of 'backend' suboption.
[vo/opengl] error creating CGL context: invalid pixel format (10002)
Error opening/initializing the selected video_out (-vo) device.
Video: no video


Exiting... (Errors when loading file)
---

Any ideas what all that means?

Greets,
Marko

Re: [macports-ports] branch master updated: fontconfig: support mpkg/mdmg packaging, add postinstall script to initialize/update font cache. Fixes #53350

2017-01-21 Thread Marko Käning 
Hi Ryan,

On 20 Jan 2017, at 05:13 , Ryan Schmidt  wrote:

> Remember that now that we've switched to GitHub, "#53350" refers to *GitHub 
> pull request* 53350.

this is why I suggested a pre-commit sanity check of commit messages in 

https://trac.macports.org/ticket/52763

I admit that it might not be so easily implementable, but still, for commits 
like this it seems to make a lot of sense, since you don’t want to mess up a 
projects’ commit history by filling it with bogus links to “PRs” which are in 
fact original MacPorts trac ticket references.

Greets,
Marko