Re: add application:start_all/1-2 (recursive OTP app starting)

Hi Siri,

I've amended the commit and rebased over maint to implement the
recommendations you had. I extended the tests and the docs, but
I didn't rename the branch, though.

git fetch git://github.com/ferd/otp.git start-all

https://github.com/ferd/otp/compare/erlang:maint...start-all
https://github.com/ferd/otp/compare/erlang:maint...start-all.patch

Regards,
Fred.

On 07/11, Siri Hansen wrote:
> Hi Fred!
> 
> OTP technical board has discussed your patch and we find the functionality
> useful. We would, however, like to see some adjustments:
> 
> 1. Since the function behaves more like application:ensure_started/1,2 than
> application:start/1,2 we would like the new name to be
> ensure_all_started/1,2.
> 
> 2. On failure, the current implementation potentially leaves one or more
> applications started. We would like it to clean up after itself in this
> situation.
> 
> 3. On success, we would like the function to return {ok,StartedApps}, where
> StartedApps is the list of applications (names) that were actually started.
> 
> Would it be possible for you to update the patch accordingly?
> 
> Thanks for your contribution!
> Regards
> /siri
> 
> 
> 
> 2013/6/24 Fredrik <[email protected]>
> 
> > On 06/20/2013 08:16 PM, Fred Hebert wrote:
> >
> >> This adds two new functions to the application module that will
> >> recursively look for missing dependencies to start the entire dependency
> >> chain of a given application, and then the application itself, if
> >> possible.
> >>
> >> Its entire behavior is based on the currently existing 'start/1-2'
> >> functions, which will report dependencies that have not yet been started
> >> and other problems encountered. 'start_all/1-2' will use this
> >> information the way any programmer would manually do it to automate the
> >> process.
> >>
> >> This kind of functionality, while not OTP-ish (releases do this start
> >> sequence for the user), is still useful for testing purposes, quick
> >> demonstrations, and programmers who do not wish or know how to use OTP
> >> releases.
> >>
> >> The function returns 'ok' for all started or already started
> >> applications, and will return '{error, {AppName, Reason}}' on a failure
> >> to allow users to rapidly know why the function failed.
> >>
> >> I decided to call it start_all even though its semantics are somewhat
> >> similar to ensure_start in how it ignores already_started applications,
> >> but am open to changing the name if the OTP team prefers
> >> 'ensure_all/1-2' as functions for this module.
> >>
> >> Documentation has been added, and so were tests, looking for a few
> >> simple case: simple app, app with a dependency, and apps with a
> >> 3rd-level dependency not booting, all locally.
> >>
> >> Note that no specific check is made for direct or indirect circular
> >> dependencies between applications, and these will make the function go
> >> in an infinite loop.
> >>
> >> git fetch git://github.com/ferd/otp.git start-all
> >>
> >> https://github.com/ferd/otp/**compare/erlang:maint...start-**all<https://github.com/ferd/otp/compare/erlang:maint...start-all>
> >> https://github.com/ferd/otp/**compare/erlang:maint...start-**all.patch<https://github.com/ferd/otp/compare/erlang:maint...start-all.patch>
> >>
> >> Regards,
> >> Fred Hebert.
> >> ______________________________**_________________
> >> erlang-patches mailing list
> >> [email protected]
> >> http://erlang.org/mailman/**listinfo/erlang-patches<http://erlang.org/mailman/listinfo/erlang-patches>
> >>
> > Hello Fred,
> > I've fetched your patch and assigned it to be reviewed.
> > Thanks,
> >
> > --
> >
> > BR Fredrik Gustafsson
> > Erlang OTP Team
> >
> >
> > ______________________________**_________________
> > erlang-patches mailing list
> > [email protected]
> > http://erlang.org/mailman/**listinfo/erlang-patches<http://erlang.org/mailman/listinfo/erlang-patches>
> >
_______________________________________________
erlang-patches mailing list
[email protected]
http://erlang.org/mailman/listinfo/erlang-patches