groovy-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mario Garcia <mario.g...@gmail.com>
Subject Re: What the... static compile by default
Date Fri, 11 May 2018 15:32:50 GMT
*About Static Compilation changes:*

I've used the way it's documented in the official documentation, and I
agree with Cedric, I don't like having a system property. I see more
benefits using the compiler configuration file:

   - Configuration is more fine grained (apply to all, apply to some
   classes, apply to some packages...)
   - All compilation configuration can be found in one place. Having more
   than one place to do this could be error prone, and harder to maintain.
   - System properties are normally used when the process should vary
   depending on the environment. In this case, I'm wondering why I would want
   to compile my code statically in one environment but dynamically in
   another. Maybe there is a case for that, but to me is weird.

*About Daniel response:*

I'm so sad to hear that Daniel. In the past few years I've been hearing
only amazing things coming from your contributions.  Like someone has
mentioned, Groovy 3 wouldn't be the same without you. I really hope you
could reconsider your decision and keep contributing to Groovy.

*About doing commits on master:*

Reading the "Contributing code" section, at groovy-lang.org it seems
everybody should be creating a local branch and to a MR afterwards over the
remote version of that local branch. So  (again, reading the
documentation)  nobody should be adding commits to master directly.

I think merge requests are essential. I'm reading Jochen is saying that
this is not very straight forward with Github. Could anyone please explain
why ? Knowing the pains may help finding the solution.

My two cents
Mario

El vie., 11 may. 2018 3:42, Thibault Kruse <tibokruse@googlemail.com>
escribió:

> It seems a bit weird to leave this thread dangling after the dramatic
> entry scene.
>
> The activity on master branch seems to indicate some changes were decided:
>
> danielsun1106 committed 2 days ago : Revert "GROOVY-8543: Support
> setting compileStatic by default via sys…
> danielsun1106 committed 18 hours ago : GROOVY-7204: Static type
> checking and compilation fail when multiple …
> danielsun1106 committed 14 hours ago : Simplify finding generics
> implementation class
>
> However, the meta-concern by Cedric was not addressed it seems. Why is
> anyone directly working on the master branch of groovy?
> Is there a technical reason for this, rather than using feature
> branches, code reviews, and merge approvals?
> Or is it just that nobody would have time to review in a timely
> fashion anyway, so it's either that or zero progress?
>
>
> On Tue, May 8, 2018 at 7:43 AM, MG <mgbiz@arscreat.com> wrote:
> > On 07.05.2018 17:54, Cédric Champeau wrote:
> >>
> >> I'd typically very much prefer a custom file extension for example.
> >
> >
> > That would be my preferred way to give anyone a simple mean to choose
> static
> > compilation as the default for a Groovy file. Afair the counter argument
> > was, that Groovy compiles any file with any extension in dynamic mode by
> > default, so this might be a breaking change if someone has used the
> picked
> > extension for his files. Groovy 3.0 might be the right spot to introduce
> > something like this, since there will be breaking changes anyway...
> >
> >> That said, since I'm not contributing code anymore (my last contribution
> >> was rewriting most of the build, which I hope was helpful),
> >
> >
> > Any improvement/speedup of the Gradle build was _definitely_ appreciated
> :-)
> >
> >> I'm happy to step down and let you work as you wish.
> >
> >
> > This is tricky: One cannot agree with just any direction someone who
> invests
> > the time to advance Groovy wants to take it too, that would be taking
> > Doocracy too far, imho, and might lead to a Groovy which is much worse
> than
> > it could be.
> > In this particular case I am torn: I think we could definitely live with
> the
> > system property, I don't feel there is a large probability that it will
> > break anything. On the other hand, using the existing mechanism, or
> > introducing a static compilation source file extension, or a compiler
> switch
> > seem to me to be the better choices - but maybe Daniel can explain why he
> > went with the property approach ?-)
> >
> > Cheers,
> > mg
> >
> >
> >
>

Mime
View raw message