Anyone opposed to removing -make and fsc?

[Simon Ochsenreither]

Not sure if we ever managed to communicate the state of fsc properly ("Use sbt/zinc/... instead"), but the -make scalac option is already marked as deprecated.

Shall we get rid of them for 2.11?

Btw, for how long do we want to keep the sbaz documentation on scala-lang.org? (http://www.scala-lang.org/node/93)

[Simon Ochsenreither]

Regarding “removing fsc”: https://github.com/soc/scala/compare/topic;remove-fsc?expand=1

Does someone know whether any third-party tools still use fsc? Afair at least IntelliJ offered an option to use fsc in the past, but afaik they are migrating/have migrated to SBT/zinc.

[Martin Grigorov]

Hi,

On Tue, Jul 2, 2013 at 2:19 PM, Simon Ochsenreither <simon.och...@gmail.com> wrote:

Regarding “removing fsc”: https://github.com/soc/scala/compare/topic;remove-fsc?expand=1

Does someone know whether any third-party tools still use fsc? Afair at least IntelliJ offered an option to use fsc in the past, but afaik they are migrating/have migrated to SBT/zinc.

If you disable "External compiler" in IDEA settings then the FSC settings reappear again in the UI.

Recently few users complained in IDEA Scala plugin forum that the external compiler (based on Zinc) is actually slower for them than FSC so they wanted to re-enable FSC.

Here is a recent discussion: http://devnet.jetbrains.com/thread/445950

 

--
You received this message because you are subscribed to the Google Groups "scala-internals" group.
To unsubscribe from this group and stop receiving emails from it, send an email to scala-interna...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Simon Ochsenreither]

If you disable "External compiler" in IDEA settings then the FSC settings reappear again in the UI.

Recently few users complained in IDEA Scala plugin forum that the external compiler (based on Zinc) is actually slower for them than FSC so they wanted to re-enable FSC.

From what I've read, IntelliJ13 will remove support for fsc anyway ... this of course means that it gets even more important to figure out in which cases zinc is slower than fsc and fix those issues.

Thanks a lot! Looks like this thread is also helpful: http://devnet.jetbrains.com/message/5491387

[Mark Harrah]

On Tue, 2 Jul 2013 04:38:35 -0700 (PDT)
Simon Ochsenreither <simon.och...@gmail.com> wrote:

>
>
> > If you disable "External compiler" in IDEA settings then the FSC settings
> > reappear again in the UI.
> > Recently few users complained in IDEA Scala plugin forum that the external
> > compiler (based on Zinc) is actually slower for them than FSC so they
> > wanted to re-enable FSC.
> >
>
> From what I've read, IntelliJ13 will remove support for fsc anyway ... this
> of course means that it gets even more important to figure out in which
> cases zinc is slower than fsc and fix those issues.

fsc keeps the resident compiler instance and zinc does not. This adds an overhead to running the compiler at each step in zinc. This is not planned to change in the near future, since a reliable resident compiler is hard. Grzegorz might have more to say on this.

> Thanks a lot! Looks like this thread is also helpful:
> http://devnet.jetbrains.com/message/5491387

I am surprised that fsc is ever slower than zinc. The benefit of zinc is that it is more reliable, not that it is faster than fsc. I would think that fsc is the limit that zinc would approach, but I forget the details of fsc beyond that it keeps the compiler resident.

-Mark

[Simon Ochsenreither]

I am surprised that fsc is ever slower than zinc.  The benefit of zinc is that it is more reliable, not that it is faster than fsc.  I would think that fsc is the limit that zinc would approach, but I forget the details of fsc beyond that it keeps the compiler resident.

Makes sense.

These days I'm just praying that some of the smaller improvements have unexpected (or should I say unrealistic) compilation performance gains (like the work on HashSet[Symbol], cutting down the compiler, removing constant inlining, the recent memory-leak fix, ...), but I'm increasingly approaching the state where I'm wondering if there are any papers out there about implementing type-checking on GPUs.

I wonder how a scalac AOT-compiled to native code will stack up against the the current status quo. I guess I will just go ahead and build a scalac binary and benchmark it as soon as I have some time.

Thanks for your input!

Bye,

Simon