Executable files in Boost git repostories

classic Classic list List threaded Threaded
23 messages Options
12
Reply | Threaded
Open this post in threaded view
|

Executable files in Boost git repostories

Boost - Dev mailing list
We have again run into the situation where files with a Linux executable
permission have been committed to various Boost git repositories, with
Jim King creating a PR and list of these files in Boost Admin. I have
fixed these for the repositories for which I have write access, and
created PRs for the other repositories. But this begs the question as to
what Boost's stance should be about adding actual executable files to a
Boost git repository ? As an example a Linux bash command file was added
to a particular repository and I created a PR to remove the executable
file permission from the file. But the maintainer of the repository
feels this is wrong and the Linux bash file should retain the executable
file permissions and that the file should be part of the repository. But
of course I am more interested here about the general principal of the
matter. Obviously operating system command/batch files are executable
files, but should they be so in a repository.

Finally should be not have some sort of git hook that somehow eliminates
this recurring problem of files with executable permissions being
periodically committed to a Boost git repository ? Most all of the files
I "fixed" or created PRs for are clearly source files what can not be
"executed" in any way. But this happening again is a real problem to
have to clean up each time.


_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
On 4/12/2018 13:24, Edward Diener wrote:

> We have again run into the situation where files with a Linux executable
> permission have been committed to various Boost git repositories, with
> Jim King creating a PR and list of these files in Boost Admin. I have
> fixed these for the repositories for which I have write access, and
> created PRs for the other repositories. But this begs the question as to
> what Boost's stance should be about adding actual executable files to a
> Boost git repository ? As an example a Linux bash command file was added
> to a particular repository and I created a PR to remove the executable
> file permission from the file. But the maintainer of the repository
> feels this is wrong and the Linux bash file should retain the executable
> file permissions and that the file should be part of the repository. But
> of course I am more interested here about the general principal of the
> matter. Obviously operating system command/batch files are executable
> files, but should they be so in a repository.

Files which are not actually executable scripts should not have that bit
set, of course.

Files which are executable scripts generally should have the bit set,
even in the repository.

However then the question becomes: are these scripts only for the
maintainer's use (in which case perhaps they shouldn't be in the
repository?) or are they intended for user use (in which case what
happens on platforms that cannot run the script?)

So for portability reasons, in my opinion, it's probably better to get
b2 to do things rather than writing custom scripts that don't work on
all platforms.

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
On 2018-12-03 19:31, Gavin Lambert via Boost wrote:

> On 4/12/2018 13:24, Edward Diener wrote:
>> We have again run into the situation where files with a Linux
>> executable permission have been committed to various Boost git
>> repositories, with Jim King creating a PR and list of these files in
>> Boost Admin. I have fixed these for the repositories for which I have
>> write access, and created PRs for the other repositories. But this
>> begs the question as to what Boost's stance should be about adding
>> actual executable files to a Boost git repository ? As an example a
>> Linux bash command file was added to a particular repository and I
>> created a PR to remove the executable file permission from the file.
>> But the maintainer of the repository feels this is wrong and the
>> Linux bash file should retain the executable file permissions and
>> that the file should be part of the repository. But of course I am
>> more interested here about the general principal of the matter.
>> Obviously operating system command/batch files are executable files,
>> but should they be so in a repository.
>
> Files which are not actually executable scripts should not have that
> bit set, of course.
>
> Files which are executable scripts generally should have the bit set,
> even in the repository.
>
> However then the question becomes: are these scripts only for the
> maintainer's use (in which case perhaps they shouldn't be in the
> repository?) or are they intended for user use (in which case what
> happens on platforms that cannot run the script?)
>
> So for portability reasons, in my opinion, it's probably better to get
> b2 to do things rather than writing custom scripts that don't work on
> all platforms.


Why do we even have to discuss this here globally, rather than trusting
project maintainers to know what they are doing ?


Stefan

--

      ...ich hab' noch einen Koffer in Berlin...
   


_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
In reply to this post by Boost - Dev mailing list
AMDG

On 12/03/2018 05:24 PM, Edward Diener via Boost wrote:

> We have again run into the situation where files with a Linux executable
> permission have been committed to various Boost git repositories, with
> Jim King creating a PR and list of these files in Boost Admin. I have
> fixed these for the repositories for which I have write access, and
> created PRs for the other repositories. But this begs the question as to
> what Boost's stance should be about adding actual executable files to a
> Boost git repository ? As an example a Linux bash command file was added
> to a particular repository and I created a PR to remove the executable
> file permission from the file. But the maintainer of the repository
> feels this is wrong and the Linux bash file should retain the executable
> file permissions and that the file should be part of the repository. But
> of course I am more interested here about the general principal of the
> matter. Obviously operating system command/batch files are executable
> files, but should they be so in a repository.
>

Yes.

> Finally should be not have some sort of git hook that somehow eliminates
> this recurring problem of files with executable permissions being
> periodically committed to a Boost git repository ? Most all of the files
> I "fixed" or created PRs for are clearly source files what can not be
> "executed" in any way. But this happening again is a real problem to
> have to clean up each time.
>

That would be reasonable as long as it doesn't accidentally
reject files that should be executable.

In Christ,
Steven Watanabe

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
In reply to this post by Boost - Dev mailing list
AMDG

On 12/03/2018 05:31 PM, Gavin Lambert via Boost wrote:

> On 4/12/2018 13:24, Edward Diener wrote:
>> <snip>Obviously
>> operating system command/batch files are executable files, but should
>> they be so in a repository.
>
> Files which are not actually executable scripts should not have that bit
> set, of course.
>
> Files which are executable scripts generally should have the bit set,
> even in the repository.
>
> However then the question becomes: are these scripts only for the
> maintainer's use (in which case perhaps they shouldn't be in the
> repository?)

Such scripts belong in the repository anyway.
Else what would we do for libraries whose
maintainers' disappear?  Or if a library
has multiple maintainers?

> or are they intended for user use (in which case what
> happens on platforms that cannot run the script?)
>

Just provide multiple scripts?

> So for portability reasons, in my opinion, it's probably better to get
> b2 to do things rather than writing custom scripts that don't work on
> all platforms.
>

I agree with Stefan.  Deciding how to handle these
scripts is the responsibility of each individual
maintainer.

In Christ,
Steven Watanabe

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
On Mon, Dec 3, 2018 at 4:52 PM Steven Watanabe via Boost
<[hidden email]> wrote:
> > However then the question becomes: are these scripts only for the
> > maintainer's use (in which case perhaps they shouldn't be in the
> > repository?)

Beast needs a couple of shell scripts with the execute bit set
otherwise they won't run correctly on Travis. For example:

<https://github.com/boostorg/beast/blob/develop/tools/build-and-test.sh>

(Note in the header above the file it says "Executable File", because
the -x bit is set).

Thanks

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
In reply to this post by Boost - Dev mailing list
On 4/12/2018 13:52, Steven Watanabe wrote:
>> However then the question becomes: are these scripts only for the
>> maintainer's use (in which case perhaps they shouldn't be in the
>> repository?)
>
> Such scripts belong in the repository anyway.
> Else what would we do for libraries whose
> maintainers' disappear?  Or if a library
> has multiple maintainers?

Depends what the scripts do, of course.  I was thinking of scripts that
used paths or packages local to the maintainer's machine and so wouldn't
work for anyone else anyway.

> Just provide multiple scripts?

For Windows and Linux, that's not too difficult.  What about other
platforms?  Especially those that the maintainer cannot test?

> I agree with Stefan.  Deciding how to handle these
> scripts is the responsibility of each individual
> maintainer.

That is true, but it's not silly to provide guidelines.

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
In reply to this post by Boost - Dev mailing list
On 12/3/2018 7:41 PM, Steven Watanabe via Boost wrote:

> AMDG
>
> On 12/03/2018 05:24 PM, Edward Diener via Boost wrote:
>> We have again run into the situation where files with a Linux executable
>> permission have been committed to various Boost git repositories, with
>> Jim King creating a PR and list of these files in Boost Admin. I have
>> fixed these for the repositories for which I have write access, and
>> created PRs for the other repositories. But this begs the question as to
>> what Boost's stance should be about adding actual executable files to a
>> Boost git repository ? As an example a Linux bash command file was added
>> to a particular repository and I created a PR to remove the executable
>> file permission from the file. But the maintainer of the repository
>> feels this is wrong and the Linux bash file should retain the executable
>> file permissions and that the file should be part of the repository. But
>> of course I am more interested here about the general principal of the
>> matter. Obviously operating system command/batch files are executable
>> files, but should they be so in a repository.
>>
>
> Yes.
>
>> Finally should be not have some sort of git hook that somehow eliminates
>> this recurring problem of files with executable permissions being
>> periodically committed to a Boost git repository ? Most all of the files
>> I "fixed" or created PRs for are clearly source files what can not be
>> "executed" in any way. But this happening again is a real problem to
>> have to clean up each time.
>>
>
> That would be reasonable as long as it doesn't accidentally
> reject files that should be executable.

Then probably the right thing to so is to establish which file types are
not executable files and then check only those file types to see if they
have the executable bit set and reject the push if any of them do.
Working the opposite way it would be much too easy erroneously
identifying a file as not executable when it might be.

>
> In Christ,
> Steven Watanabe
>
> _______________________________________________
> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
>



_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
In reply to this post by Boost - Dev mailing list
On 12/4/18 3:24 AM, Edward Diener via Boost wrote:

> We have again run into the situation where files with a Linux executable
> permission have been committed to various Boost git repositories, with
> Jim King creating a PR and list of these files in Boost Admin. I have
> fixed these for the repositories for which I have write access, and
> created PRs for the other repositories. But this begs the question as to
> what Boost's stance should be about adding actual executable files to a
> Boost git repository ? As an example a Linux bash command file was added
> to a particular repository and I created a PR to remove the executable
> file permission from the file. But the maintainer of the repository
> feels this is wrong and the Linux bash file should retain the executable
> file permissions and that the file should be part of the repository. But
> of course I am more interested here about the general principal of the
> matter. Obviously operating system command/batch files are executable
> files, but should they be so in a repository.

Yes, script files that are supposed to be run should have the executable
bit set in git as well. The problem is when this bit is set for files
that are not supposed to be executable (e.g. C++ source files or docs).
These instances need to be fixed and, preferably, prevented from
appearing in the future.

> Finally should be not have some sort of git hook that somehow eliminates
> this recurring problem of files with executable permissions being
> periodically committed to a Boost git repository ? Most all of the files
> I "fixed" or created PRs for are clearly source files what can not be
> "executed" in any way. But this happening again is a real problem to
> have to clean up each time.

Yes, I'm in favor of adding some sort of a server-side git hook. For
that all script files need to have an appropriate extension so that they
can be whitelisted to have an executable permission. Though I'm not sure
how server-side git hooks work with GitHub.

A separate concern of mine is binary executables. IMHO, those should be
outright banned, executable or not. Not that I remember seeing ones in
git, but while we're at it, we could enforce it as well.

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
In reply to this post by Boost - Dev mailing list
On 12/4/18 3:40 AM, Stefan Seefeld via Boost wrote:
>
> Why do we even have to discuss this here globally, rather than trusting
> project maintainers to know what they are doing ?

Apparently, maintainers can't be trusted as this issue comes up on and
on again.

Look at this as a sort of automated testing - you shouldn't be able to
push bogus or straight dangerous stuff because you're affecting all git
users.

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
In reply to this post by Boost - Dev mailing list
On 12/4/18 3:31 AM, Gavin Lambert via Boost wrote:
>
> However then the question becomes: are these scripts only for the
> maintainer's use (in which case perhaps they shouldn't be in the
> repository?) or are they intended for user use (in which case what
> happens on platforms that cannot run the script?)
>
> So for portability reasons, in my opinion, it's probably better to get
> b2 to do things rather than writing custom scripts that don't work on
> all platforms.

I agree with you but only as long as the script is supposed to be run by
users, and routinely so. There's no reason to limit yourself to
Boost.Build otherwise. Remember, Boost.Build is not a general scripting
language, it's a build system. It's not as flexible or convenient as a
shell script in what a shell script can do, provided that you're in a
POSIX environment. Let alone, Python, for example.

And before anyone asks, if you don't have a POSIX environment, you
should really get one. Even on Windows it shouldn't be much of a
problem. It will do you good beyond Boost as well.

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
In reply to this post by Boost - Dev mailing list
On Tue, Dec 4, 2018 at 4:03 AM Andrey Semashev via Boost <
[hidden email]> wrote:

>
> Yes, I'm in favor of adding some sort of a server-side git hook. For
> that all script files need to have an appropriate extension so that they
> can be whitelisted to have an executable permission. Though I'm not sure
> how server-side git hooks work with GitHub.
>

I don't think there's anything like git server-side hooks. Best you can do
is write a GitHub Application. As otherwise it's all client side hooks.
Other option is to add these checks to the regular library conformance
testing <
https://github.com/boostorg/boost/blob/master/status/boost_check_library.py
>.

--
-- Rene Rivera
-- Grafik - Don't Assume Anything
-- Robot Dreams - http://robot-dreams.net

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
On 12/4/2018 8:03 AM, Rene Rivera via Boost wrote:

> On Tue, Dec 4, 2018 at 4:03 AM Andrey Semashev via Boost <
> [hidden email]> wrote:
>
>>
>> Yes, I'm in favor of adding some sort of a server-side git hook. For
>> that all script files need to have an appropriate extension so that they
>> can be whitelisted to have an executable permission. Though I'm not sure
>> how server-side git hooks work with GitHub.
>>
>
> I don't think there's anything like git server-side hooks.

https://git-scm.com/docs/githooks and
https://git-scm.com/book/en/v2/Customizing-Git-Git-Hooks suggest
otherwise, although understanding how to write a hook from the sparse
documentation seems difficult to me. Maybe someone who has written a git
hook before understands how we can cycle through the files in a 'git
push' and, if one of those files is a source file with its executable
bit set, reject the push.

> Best you can do
> is write a GitHub Application. As otherwise it's all client side hooks.
> Other option is to add these checks to the regular library conformance
> testing <
> https://github.com/boostorg/boost/blob/master/status/boost_check_library.py
>> .
>



_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
On Tue, Dec 4, 2018 at 9:28 AM Edward Diener via Boost <
[hidden email]> wrote:

> On 12/4/2018 8:03 AM, Rene Rivera via Boost wrote:
> > On Tue, Dec 4, 2018 at 4:03 AM Andrey Semashev via Boost <
> > [hidden email]> wrote:
> >
> >>
> >> Yes, I'm in favor of adding some sort of a server-side git hook. For
> >> that all script files need to have an appropriate extension so that they
> >> can be whitelisted to have an executable permission. Though I'm not sure
> >> how server-side git hooks work with GitHub.
> >>
> >
> > I don't think there's anything like git server-side hooks.
>
> https://git-scm.com/docs/githooks and
> https://git-scm.com/book/en/v2/Customizing-Git-Git-Hooks suggest
> otherwise, although understanding how to write a hook from the sparse
> documentation seems difficult to me. Maybe someone who has written a git
> hook before understands how we can cycle through the files in a 'git
> push' and, if one of those files is a source file with its executable
> bit set, reject the push.
>

GitHub doesn't allow those server side hooks.

--
-- Rene Rivera
-- Grafik - Don't Assume Anything
-- Robot Dreams - http://robot-dreams.net

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
In reply to this post by Boost - Dev mailing list
On 2018-12-04 5:06 a.m., Andrey Semashev via Boost wrote:

> On 12/4/18 3:40 AM, Stefan Seefeld via Boost wrote:
>>
>> Why do we even have to discuss this here globally, rather than trusting
>> project maintainers to know what they are doing ?
>
> Apparently, maintainers can't be trusted as this issue comes up on and
> on again.

Take it up with those maintainers, but don't attempt to overrule them.
That's precisely the wrong direction to take.

> Look at this as a sort of automated testing - you shouldn't be able to
> push bogus or straight dangerous stuff because you're affecting all
> git users.

This is exactly the kind of thinking that annoys me with Boost. You are
trying to solve the wrong problem ! We need more autonomy, not less !

Don't patronize maintainers by imposing artificial (and often misguided)
rules.

Instead, empower boost maintainers to take full responsibility of their
projects, then establish criteria for projects to be boost members, and
drop projects that don't meet these requirements.


Stefan

--

       ...ich hab' noch einen Koffer in Berlin...
     


_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
On 12/4/18 6:41 PM, stefan via Boost wrote:

> On 2018-12-04 5:06 a.m., Andrey Semashev via Boost wrote:
>
>> On 12/4/18 3:40 AM, Stefan Seefeld via Boost wrote:
>>>
>>> Why do we even have to discuss this here globally, rather than trusting
>>> project maintainers to know what they are doing ?
>>
>> Apparently, maintainers can't be trusted as this issue comes up on and
>> on again.
>
> Take it up with those maintainers, but don't attempt to overrule them.
> That's precisely the wrong direction to take.
>
>> Look at this as a sort of automated testing - you shouldn't be able to
>> push bogus or straight dangerous stuff because you're affecting all
>> git users.
>
> This is exactly the kind of thinking that annoys me with Boost. You are
> trying to solve the wrong problem ! We need more autonomy, not less !
>
> Don't patronize maintainers by imposing artificial (and often misguided)
> rules.
>
> Instead, empower boost maintainers to take full responsibility of their
> projects, then establish criteria for projects to be boost members, and
> drop projects that don't meet these requirements.

I disagree. I really don't want to go into the debate about autonomy vs.
moderation, I'll just say that from my point of view a certain level of
moderation and uniformity is needed. If a library wants to do everything
in its own way, not caring to be part of the Boost ecosystem, it might
as well be maintained as a standalone project.

And this particular issue is not about infringing on anyone's rights,
really. Unless you want the right to shoot yourself and everyone else in
the foot. Marking files as executable, when they are not actual scripts,
is never the right thing to do.

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list

On 2018-12-04 10:54 a.m., Andrey Semashev via Boost wrote:

> On 12/4/18 6:41 PM, stefan via Boost wrote:
>> Don't patronize maintainers by imposing artificial (and often
>> misguided) rules.
>>
>> Instead, empower boost maintainers to take full responsibility of
>> their projects, then establish criteria for projects to be boost
>> members, and drop projects that don't meet these requirements.
>
> I disagree. I really don't want to go into the debate about autonomy
> vs. moderation, I'll just say that from my point of view a certain
> level of moderation and uniformity is needed. If a library wants to do
> everything in its own way, not caring to be part of the Boost
> ecosystem, it might as well be maintained as a standalone project.

I don't disagree. In fact, I believe that's what I just said: establish
criteria for adherence, and judge projects by those. But the discussion
(or at least, my argument) isn't about the right level of uniformity.
It's about patronizing (and infantilizing) project maintainers.


Stefan

--

       ...ich hab' noch einen Koffer in Berlin...
     


_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
In reply to this post by Boost - Dev mailing list
On 12/4/2018 10:34 AM, Rene Rivera via Boost wrote:

> On Tue, Dec 4, 2018 at 9:28 AM Edward Diener via Boost <
> [hidden email]> wrote:
>
>> On 12/4/2018 8:03 AM, Rene Rivera via Boost wrote:
>>> On Tue, Dec 4, 2018 at 4:03 AM Andrey Semashev via Boost <
>>> [hidden email]> wrote:
>>>
>>>>
>>>> Yes, I'm in favor of adding some sort of a server-side git hook. For
>>>> that all script files need to have an appropriate extension so that they
>>>> can be whitelisted to have an executable permission. Though I'm not sure
>>>> how server-side git hooks work with GitHub.
>>>>
>>>
>>> I don't think there's anything like git server-side hooks.
>>
>> https://git-scm.com/docs/githooks and
>> https://git-scm.com/book/en/v2/Customizing-Git-Git-Hooks suggest
>> otherwise, although understanding how to write a hook from the sparse
>> documentation seems difficult to me. Maybe someone who has written a git
>> hook before understands how we can cycle through the files in a 'git
>> push' and, if one of those files is a source file with its executable
>> bit set, reject the push.
>>
>
> GitHub doesn't allow those server side hooks.

Are you saying that we have no access to the .git or .git/hooks
subdirectory of a Github hosted repository ?


_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
On Wed, Dec 5, 2018 at 3:50 PM Edward Diener via Boost <
[hidden email]> wrote:

> On 12/4/2018 10:34 AM, Rene Rivera via Boost wrote:
> > GitHub doesn't allow those server side hooks.
>
> Are you saying that we have no access to the .git or .git/hooks
> subdirectory of a Github hosted repository ?
>

GitHub doesn't allow running scripts on their servers.

--
-- Rene Rivera
-- Grafik - Don't Assume Anything
-- Robot Dreams - http://robot-dreams.net

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Executable files in Boost git repostories

Boost - Dev mailing list
On 6/12/2018 10:53, Rene Rivera wrote:
> On Wed, Dec 5, 2018 at 3:50 PM Edward Diener wrote:
>> Are you saying that we have no access to the .git or .git/hooks
>> subdirectory of a Github hosted repository ?
>
> GitHub doesn't allow running scripts on their servers.

You can write a webhook to run a script on a different server.  But at
best this could yell at someone after the fact; it can't prevent
accepting a push (although it can mark a pull request as invalid --
that's how the CI hooks in).

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
12