Scala 2.9.1 RC1

On Mon, Jul 25, 2011 at 4:37 PM, Antonio Cunei wrote:
> We are happy to announce the first release candidate of the new
> Scala distribution: Scala 2.9.1 RC1 is now available from our
> In addition, the following changes (bug fixes and improvements)
> have also been ported:
>
> Among the bug fixed in this release are:

A few more for consideration:

https://issues.scala-lang.org/browse/SI-4752
Hash code for None changes across JVM invocations
Fixed (r25215)

https://issues.scala-lang.org/browse/SI-4782
scalac ignores empty classpath argument
Open, Regression

-jason

On Mon, Jul 25, 2011 at 4:37 PM, Antonio Cunei wrote:
> Please help us testing this new release candidate, and let us know
> of any issues you may detect, or of any binary incompatibility
> issue that you may encounter. A description of the improvements in
> Scala 2.9.1 is available below.

sbt 0.7.7 isn't working with 2.9.1.RC1 probably because 25068 is missing.

> Please help us testing this new release candidate, and let us know
> of any issues you may detect, or of any binary incompatibility
> issue that you may encounter. A description of the improvements in
> Scala 2.9.1 is available below.

When I run the interpreter then directly type :q to quit, the
interpreter hangs. If I type any expression before :q, the interpreter
stops correctly.

On Mon, 25 Jul 2011 16:57:21 +0200
Jason Zaugg wrote:

> On Mon, Jul 25, 2011 at 4:37 PM, Antonio Cunei wrote:
> > We are happy to announce the first release candidate of the new
> > Scala distribution: Scala 2.9.1 RC1 is now available from our
> > In addition, the following changes (bug fixes and improvements)
> > have also been ported:
> >
> > Among the bug fixed in this release are:
>
> A few more for consideration:
>
> https://issues.scala-lang.org/browse/SI-4752
> Hash code for None changes across JVM invocations
> Fixed (r25215)
>
> https://issues.scala-lang.org/browse/SI-4782
> scalac ignores empty classpath argument
> Open, Regression

https://issues.scala-lang.org/browse/SI-4757
Nonexisting symbol error with separate compilation

If fixes for these three could be included, I believe sbt would move to 2.9.1 now that the performance issues have been resolved.

-Mark

I wonder if this is the same as https://issues.scala-lang.org/browse/SI-4705

You could run jstack and see if you get a similar thread dump to the
one I attached. If it looks similar the issue should be reopened.

--Jean-Laurent

On Jul 26, 1:42 am, ramzi wrote:
> > Please help us testing this new release candidate, and let us know
> > of any issues you may detect, or of any binary incompatibility
> > issue that you may encounter. A description of the improvements in
> > Scala 2.9.1 is available below.
>
> When I run the interpreter then directly type :q to quit, the
> interpreter hangs. If I type any expression before :q, the interpreter
> stops correctly.
>

On Tue, Jul 26, 2011 at 2:16 PM, Mark Harrah wrote:
> I believe sbt would move to 2.9.1 now that the performance issues have been resolved.

Have you had a chance to test SBT with 2.9.1 RC1 to verify that
performance is as good or better than 2.8.1?

Best,
Ismael

Our project doesn't compile for 2.8.1, so here are the numbers for
2.9.1.RC1 and 2.9.0-1:

sbt '++2.9.0-1 test-compile' 284.73s user 4.06s system 293% cpu 1:38.25 total
sbt '++2.9.0-1 test-compile' 283.97s user 4.65s system 285% cpu 1:41.25 total

sbt '++2.9.1.RC1 test-compile' 237.86s user 2.89s system 299% cpu 1:20.27 total
sbt '++2.9.1.RC1 test-compile' 234.71s user 2.70s system 293% cpu 1:21.00 total

So it's 100 seconds for 2.9.0 vs 80 seconds in 2.9.1.RC1.

On Tue, Jul 26, 2011 at 4:19 PM, Ismael Juma wrote:
> On Tue, Jul 26, 2011 at 2:16 PM, Mark Harrah wrote:
>> I believe sbt would move to 2.9.1 now that the performance issues have been resolved.
>
> Have you had a chance to test SBT with 2.9.1 RC1 to verify that
> performance is as good or better than 2.8.1?
>
> Best,
> Ismael
>

On 7/26/11 6:16 AM, Mark Harrah wrote:
>> https://issues.scala-lang.org/browse/SI-4752 Hash code for None
>> changes across JVM invocations Fixed (r25215)
>>
>> https://issues.scala-lang.org/browse/SI-4782 scalac ignores empty
>> classpath argument Open, Regression
>
> https://issues.scala-lang.org/browse/SI-4757 Nonexisting symbol error
> with separate compilation
>
> If fixes for these three could be included, I believe sbt would move
> to 2.9.1 now that the performance issues have been resolved.

These are all backported to 2.9.x (and the not-yet-fixed one fixed in
trunk.) After I reverify tests/binary-compat we'll roll an RC2.

On Tue, Jul 26, 2011 at 6:05 PM, Paul Phillips <paulp@improving.org> wrote:

On 7/26/11 6:16 AM, Mark Harrah wrote:

https://issues.scala-lang.org/browse/SI-4752 Hash code for None
changes across JVM invocations Fixed (r25215)

https://issues.scala-lang.org/browse/SI-4782 scalac ignores empty
classpath argument Open, Regression

https://issues.scala-lang.org/browse/SI-4757 Nonexisting symbol error
with separate compilation

If fixes for these three could be included, I believe sbt would move
to 2.9.1 now that the performance issues have been resolved.

These are all backported to 2.9.x (and the not-yet-fixed one fixed in trunk.) After I reverify tests/binary-compat we'll roll an RC2.

I owe you a cookie or 10.

--
Viktor Klang

Akka Tech LeadTypesafe - Enterprise-Grade Scala from the Experts

Twitter: @viktorklang

On Tue, Jul 26, 2011 at 5:05 PM, Paul Phillips wrote:
> These are all backported to 2.9.x (and the not-yet-fixed one fixed in
> trunk.) After I reverify tests/binary-compat we'll roll an RC2.

Great to see such quick turnaround. By the way, I've tested RC1 and
all of our tests passed.

Best,
Ismael

On 7/26/11 7:19 AM, huynhjl wrote:
> I wonder if this is the same as https://issues.scala-lang.org/browse/SI-4705
>
> You could run jstack and see if you get a similar thread dump to the
> one I attached. If it looks similar the issue should be reopened.

I realize jira makes it comically difficult to tell whether an issue is
open, which itself is a ticket:

https://issues.scala-lang.org/browse/INF-3

But SI-4705 has been reopened since jun 18th. I'm not overlooking it, I
just don't yet know what is the cause.

On Tue, 26 Jul 2011 15:19:52 +0100
Ismael Juma wrote:

> On Tue, Jul 26, 2011 at 2:16 PM, Mark Harrah wrote:
> > I believe sbt would move to 2.9.1 now that the performance issues have been resolved.
>
> Have you had a chance to test SBT with 2.9.1 RC1 to verify that
> performance is as good or better than 2.8.1?

Yes. It is about the same as 2.8.1, perhaps a bit faster in some cases.

-Mark

On Wed, Jul 27, 2011 at 3:28 AM, Mark Harrah wrote:
> Yes.  It is about the same as 2.8.1, perhaps a bit faster in some cases.

Excellent.

Ismael

Any chance to fix #4846 [1]?

It seems that since the great getClass fix, the getClass method leaks
into type signatures resulting in things like:

scala> List(1) ++ List('a')
res0: List[AnyVal{def getClass(): java.lang.Class[_>: Int with Char<: AnyVal]}] = List(1, a)

Thanks and bye!

Simon

[1] https://issues.scala-lang.org/browse/SI-4846