Nightlies pushed to wrong directory

On 3 February 2012 11:48, iulian dragos <jaguarul@gmail.com> wrote:

The nightlies for the past two days are pushed to the following directory:
http://scala-tools.org/repo-snapshots/org/scala-lang/scala-compiler/$%257Bversion.major%257D.$%257Bversion.minor%257D.$%257Bversion.patch%257D-SNAPSHOT/

It looks like some variables are not expanded correctly. Joys of untyped programming! :)

I discovered this as well. It's caused by fbd5efe49cf23b446762dfa5026e8bac82ab04fc. I might be missing something but I don't understand why build.properties was causing troubles. If we decide this is a right way to move forward publishing nightlies should be easy to fix (you need look into src/maven).
--
Grzegorz Kossakowski

On 3 February 2012 13:34, Grzegorz Kossakowski <grzegorz.kossakowski@gmail.com> wrote:

I discovered this as well. It's caused by fbd5efe49cf23b446762dfa5026e8bac82ab04fc. I might be missing something but I don't understand why build.properties was causing troubles. If we decide this is a right way to move forward publishing nightlies should be easy to fix (you need look into src/maven).

Actually, the file is src/build/maven/pack.xml and it's not that easy to fix as I thought because it really needs those version.major, version.minor, version.patch variables to construct proper Maven version.

--
Grzegorz Kossakowski

Greg, thanks for looking into this. 

On Fri, Feb 3, 2012 at 1:43 PM, Grzegorz Kossakowski <grzegorz.kossakowski@gmail.com> wrote:

On 3 February 2012 13:34, Grzegorz Kossakowski <grzegorz.kossakowski@gmail.com> wrote:

I discovered this as well. It's caused by fbd5efe49cf23b446762dfa5026e8bac82ab04fc. I might be missing something but I don't understand why build.properties was causing troubles. If we decide this is a right way to move forward publishing nightlies should be easy to fix (you need look into src/maven).

Actually, the file is src/build/maven/pack.xml and it's not that easy to fix as I thought because it really needs those version.major, version.minor, version.patch variables to construct proper Maven version.

--
Grzegorz Kossakowski

--
« Je déteste la montagne, ça cache le paysage »
Alphonse Allais

On 3 February 2012 13:52, iulian dragos <jaguarul@gmail.com> wrote:

Greg, thanks for looking into this. 

Sure. Also, since we are going through next iteration over build numbers I should mention needs of projects like Scala virtualized or Scala+GWT where we want to put a suffix after patch version that disambiguates snapshots of those project from main Scala. E.g. version numbers used by those projects are:
2.10.0-virtualized-SNAPSHOT2.10.0-scalagwt-SNAPSHOT

--
Grzegorz Kossakowski

On Fri, Feb 3, 2012 at 4:34 AM, Grzegorz Kossakowski
wrote:
> I discovered this as well. It's caused
> by fbd5efe49cf23b446762dfa5026e8bac82ab04fc. I might be missing something
> but I don't understand why build.properties was causing troubles.

The manual construction in ant was competing with git-describe to
create the version number (there's a period there where they look like
v2.10.0-v2.10.0-...) so in order to use git describe I had to remove
it from ant. I don't like leaving unused misleading stuff lying
around to leave the next guy trying to figure out what if anything it
does (having been "the next guy" most of the time, and the answer
usually is "it does nothing, but nobody ever removes anything") and I
searched the source tree and there was no direct mention of the file,
so I removed it too. I realized it could be being used in a way which
didn't mention the file, as it seems it was, but I didn't think to
search for the property names.

On 3 February 2012 15:26, Paul Phillips <paulp@improving.org> wrote:

On Fri, Feb 3, 2012 at 4:34 AM, Grzegorz Kossakowski
<grzegorz.kossakowski@gmail.com> wrote:
> I discovered this as well. It's caused
> by fbd5efe49cf23b446762dfa5026e8bac82ab04fc. I might be missing something
> but I don't understand why build.properties was causing troubles.

The manual construction in ant was competing with git-describe to
create the version number (there's a period there where they look like
v2.10.0-v2.10.0-...) so in order to use git describe I had to remove
it from ant.

Ah, ok. Now it makes perfect sense. That was the bit I was missing in the commit message :-)

 I don't like leaving unused misleading stuff lying
around to leave the next guy trying to figure out what if anything it
does (having been "the next guy" most of the time, and the answer
usually is "it does nothing, but nobody ever removes anything") and I
searched the source tree and there was no direct mention of the file,
so I removed it too.  I realized it could be being used in a way which
didn't mention the file, as it seems it was, but I didn't think to
search for the property names.

Sure, no worries. Now the most important question is: how we can make everything working again?
Parsing output of git describe output seems like the only viable option...

--
Grzegorz Kossakowski

On Fri, Feb 3, 2012 at 6:34 AM, Grzegorz Kossakowski
wrote:
> Sure, no worries. Now the most important question is: how we can make
> everything working again?

Well, I can put the file back. Is that not enough?

I did put it back, since it seems unlikely to hurt anything. Its
contents seem suspicious though.

version.minor=10
version.patch=0
version.suffix=alpha
version.major=2

Is that "alpha" used somewhere? I'm guessing it is discarded.

On Fri, Feb 3, 2012 at 4:08 PM, Paul Phillips <paulp@improving.org> wrote:

I did put it back, since it seems unlikely to hurt anything.  Its
contents seem suspicious though.

version.minor=10
version.patch=0
version.suffix=alpha
version.major=2

Is that "alpha" used somewhere? I'm guessing it is discarded.

last time I looked it wasn't used -- I restored its glamorous role in scala-virtualized's build, but figured SBT and friends will soon make it obsolete anyway

On 3 February 2012 16:19, Adriaan Moors <adriaan.moors@epfl.ch> wrote:

last time I looked it wasn't used -- I restored its glamorous role in scala-virtualized's build, but figured SBT and friends will soon make it obsolete anyway

Is sbt anywhere soon hitting master? Indeed, it has a chance to make those things easier to cope with.
--
Grzegorz Kossakowski

On Fri, Feb 3, 2012 at 7:25 AM, Grzegorz Kossakowski
wrote:
> Is sbt anywhere soon hitting master? Indeed, it has a chance to make those
> things easier to cope with.

It hit master a while ago, but the majority of the time one of scala,
sbt, or the build has offered breakage. However checking right now I
find I can build with a 0.11.3 snapshot.

I pushed a branch which might make it slightly easier.

https://github.com/paulp/scala/tree/xsbt-tweaks

% sbt +compile
Detected sbt version 0.11.3-SNAPSHOT
sbt snapshot is 0.11.3-20120130-052105
Using /Users/paulp/.sbt/0.11.3-SNAPSHOT as sbt dir, -sbt-dir to override.
[info] Loading project definition from /scratch/trunk1/project/project
[info] Loading project definition from
/Users/paulp/.sbt/0.11.3-SNAPSHOT/staging/eff7d7476f58ab859d23/project
[info] Loading project definition from /scratch/trunk1/project
[warn] Credentials file /Users/paulp/.ivy2/.credentials does not exist
[info] Set current project to scala (in build file:/scratch/trunk1/)
[info] Set current project to scala (in build file:/scratch/trunk1/)
Setting version to 2.9.1
[info] Set current project to scala (in build file:/scratch/trunk1/)
[info] Pulling [/scratch/trunk1/lib/scala-library.jar] from local cache
[info] Pulling [/scratch/trunk1/lib/scala-compiler.jar] from local cache
[info] Updating {file:/scratch/trunk1/}locker-library...
[info] Updating {file:/scratch/trunk1/}locker-compiler...
[info] Done updating.
[info] Resolving org.apache.ant#ant;1.8.2 ...
[info] Compiling 691 Scala sources and 25 Java sources to
/scratch/trunk1/target/locker-library/classes...
[info] Resolving org.apache.ant#ant-launcher;1.8.2 ...
[info] Resolving org.scala-tools.sbt#compiler-interface;0.11.3-SNAPSHOT ...
[info] Resolving org.scala-tools.sbt#interface;0.11.3-SNAPSHOT ...
[info] Resolving org.scala-lang#scala-library;2.9.1 ...
[info] Resolving jline#jline;1.0 ...
[info] Done updating.

On a related note, ScalaDoc nightlies have vanished since a few weeks, too.

On Fri, Feb 3, 2012 at 5:51 PM, Simon Ochsenreither <simon.ochsenreither@googlemail.com> wrote:

On a related note, ScalaDoc nightlies have vanished since a few weeks, too.

Are you referring to the compiler API? The library API is available and up-to-date: http://www.scala-lang.org/archives/downloads/distrib/files/nightly/docs/library/index.html

Cheers,
Vlad

As long as the major/minor/bugfix versions are in ANT memory properties, they'll get written to a build.properties file for the maven scripts to push them correctly.
Note:  The version we show in our scala.properties does not, and has never lined up with the published version.  I'm not sure what the motivation was for removing them, since as far as i know, they should not have been mucking with any versions in the build.
This is yet another reason I'd like to finally kick ANT and move to SBT.  yes, I know that in a land where SBT can't compile against trunk, it's impossible to do so.  However, these kinds of issues are handled far easier and more elegantly in SBT.  I'm over here making MSI packages assuming everything is going swimmingly, but instead we have 2 and a half build systems to maintain, and I'm pretty sure the SBT build is not using the new versioning strategy.  We really need to move to one or the other.  For the upcoming 2.10 release, I need to know what input to expect to build the native packages, and there's subtle differences in what ANT does vs. SBT.
- Josh

On Fri, Feb 3, 2012 at 9:46 AM, Paul Phillips <paulp@improving.org> wrote:

On Fri, Feb 3, 2012 at 6:34 AM, Grzegorz Kossakowski
<grzegorz.kossakowski@gmail.com> wrote:
> Sure, no worries. Now the most important question is: how we can make
> everything working again?

Well, I can put the file back.  Is that not enough?

Hi Vlad,

yes, that's the URL the nightlies were published to for years. Since a few weeks it doesn't point to the trunk ScalaDocs anymore, but to 2.9.2 instead. No idea why, considering that API changes between 2.9.1 and 2.9.2 might be non-existent to negligible. Any chance to fix it and point it to trunk again?

Thanks,

Simon

On Fri, Feb 3, 2012 at 15:05, Vlad Ureche wrote:
>
>
> On Fri, Feb 3, 2012 at 5:51 PM, Simon Ochsenreither
> wrote:
>>
>> On a related note, ScalaDoc nightlies have vanished since a few weeks,
>> too.
>
>
> Are you referring to the compiler API? The library API is available and
> up-to-date:
> http://www.scala-lang.org/archives/downloads/distrib/files/nightly/docs/...

That's 2.9.x's API, not master's.

On Fri, Feb 3, 2012 at 9:07 PM, Daniel Sobral <dcsobral@gmail.com> wrote:

On Fri, Feb 3, 2012 at 15:05, Vlad Ureche <vlad.ureche@gmail.com> wrote:
>
> Are you referring to the compiler API? The library API is available and
> up-to-date:
> http://www.scala-lang.org/archives/downloads/distrib/files/nightly/docs/library/index.html

That's 2.9.x's API, not master's.

You are right, the 2.9.2 jenkins job was replacing the trunk scaladoc pages.
Issued a pull request which should prevent any other version except trunk from publishing the scaladoc pages.

Thanks for reporting it!

Cheers,
Vlad

Sorry to be pesky about it, but it's been a week now since the nightlies get pushed in the wrong place, including last night. 

On Fri, Feb 3, 2012 at 11:11 PM, Vlad Ureche <vlad.ureche@gmail.com> wrote:

On Fri, Feb 3, 2012 at 9:07 PM, Daniel Sobral <dcsobral@gmail.com> wrote:

On Fri, Feb 3, 2012 at 15:05, Vlad Ureche <vlad.ureche@gmail.com> wrote:
>
> Are you referring to the compiler API? The library API is available and
> up-to-date:
> http://www.scala-lang.org/archives/downloads/distrib/files/nightly/docs/library/index.html

That's 2.9.x's API, not master's.

You are right, the 2.9.2 jenkins job was replacing the trunk scaladoc pages.
Issued a pull request which should prevent any other version except trunk from publishing the scaladoc pages.

Thanks for reporting it!

Cheers,
Vlad

--
« Je déteste la montagne, ça cache le paysage »
Alphonse Allais

fixed and locally-built snapshot uploaded -- I'll submit a pull request to fix src/build/pack.xml

On Tue, Feb 7, 2012 at 10:32 AM, iulian dragos <jaguarul@gmail.com> wrote:

Sorry to be pesky about it, but it's been a week now since the nightlies get pushed in the wrong place, including last night. 

On Fri, Feb 3, 2012 at 11:11 PM, Vlad Ureche <vlad.ureche@gmail.com> wrote:

On Fri, Feb 3, 2012 at 9:07 PM, Daniel Sobral <dcsobral@gmail.com> wrote:

On Fri, Feb 3, 2012 at 15:05, Vlad Ureche <vlad.ureche@gmail.com> wrote:
>
> Are you referring to the compiler API? The library API is available and
> up-to-date:
> http://www.scala-lang.org/archives/downloads/distrib/files/nightly/docs/library/index.html

That's 2.9.x's API, not master's.

You are right, the 2.9.2 jenkins job was replacing the trunk scaladoc pages.
Issued a pull request which should prevent any other version except trunk from publishing the scaladoc pages.

Thanks for reporting it!

Cheers,
Vlad

--
« Je déteste la montagne, ça cache le paysage »
Alphonse Allais

reminds me that me might want to have somebody with commit rights here in our timezone. that specific is not critical, but some day there might be one.

On Tuesday, 7. February 2012 at 11:10, Adriaan Moors wrote:

fixed and locally-built snapshot uploaded -- I'll submit a pull request to fix src/build/pack.xml

On Tue, Feb 7, 2012 at 10:32 AM, iulian dragos <jaguarul@gmail.com> wrote:

Sorry to be pesky about it, but it's been a week now since the nightlies get pushed in the wrong place, including last night. 

On Fri, Feb 3, 2012 at 11:11 PM, Vlad Ureche <vlad.ureche@gmail.com> wrote:

On Fri, Feb 3, 2012 at 9:07 PM, Daniel Sobral <dcsobral@gmail.com> wrote:

On Fri, Feb 3, 2012 at 15:05, Vlad Ureche <vlad.ureche@gmail.com> wrote:
>
> Are you referring to the compiler API? The library API is available and
> up-to-date:
> http://www.scala-lang.org/archives/downloads/distrib/files/nightly/docs/library/index.html

That's 2.9.x's API, not master's.

You are right, the 2.9.2 jenkins job was replacing the trunk scaladoc pages.
Issued a pull request which should prevent any other version except trunk from publishing the scaladoc pages.

Thanks for reporting it!

Cheers,
Vlad

--
« Je déteste la montagne, ça cache le paysage »
Alphonse Allais

Thanks for your fix! Now the nightly is back to where it belongs!

On Tue, Feb 7, 2012 at 1:40 PM, Lukas Rytz <lukas.rytz@epfl.ch> wrote:

reminds me that me might want to have somebody with commit rights here in our timezone. that specific is not critical, but some day there might be one.

On Tuesday, 7. February 2012 at 11:10, Adriaan Moors wrote:

fixed and locally-built snapshot uploaded -- I'll submit a pull request to fix src/build/pack.xml

On Tue, Feb 7, 2012 at 10:32 AM, iulian dragos <jaguarul@gmail.com> wrote:

Sorry to be pesky about it, but it's been a week now since the nightlies get pushed in the wrong place, including last night. 

On Fri, Feb 3, 2012 at 11:11 PM, Vlad Ureche <vlad.ureche@gmail.com> wrote:

On Fri, Feb 3, 2012 at 9:07 PM, Daniel Sobral <dcsobral@gmail.com> wrote:

On Fri, Feb 3, 2012 at 15:05, Vlad Ureche <vlad.ureche@gmail.com> wrote:
>
> Are you referring to the compiler API? The library API is available and
> up-to-date:
> http://www.scala-lang.org/archives/downloads/distrib/files/nightly/docs/library/index.html

That's 2.9.x's API, not master's.

You are right, the 2.9.2 jenkins job was replacing the trunk scaladoc pages.
Issued a pull request which should prevent any other version except trunk from publishing the scaladoc pages.

Thanks for reporting it!

Cheers,
Vlad

--
« Je déteste la montagne, ça cache le paysage »
Alphonse Allais

--
« Je déteste la montagne, ça cache le paysage »
Alphonse Allais

On Wed, Feb 8, 2012 at 11:18 AM, iulian dragos <jaguarul@gmail.com> wrote:

Thanks for your fix! Now the nightly is back to where it belongs!

Sorry, I was too fast. I see there are two competing pull requests, none of which is merged yet. Someone (other than Mr. Jenkins) has built and pushed to scala-tools, and that's what I saw earlier..
iulian 

On Tue, Feb 7, 2012 at 1:40 PM, Lukas Rytz <lukas.rytz@epfl.ch> wrote:

reminds me that me might want to have somebody with commit rights here in our timezone. that specific is not critical, but some day there might be one.

On Tuesday, 7. February 2012 at 11:10, Adriaan Moors wrote:

fixed and locally-built snapshot uploaded -- I'll submit a pull request to fix src/build/pack.xml

On Tue, Feb 7, 2012 at 10:32 AM, iulian dragos <jaguarul@gmail.com> wrote:

Sorry to be pesky about it, but it's been a week now since the nightlies get pushed in the wrong place, including last night. 

On Fri, Feb 3, 2012 at 11:11 PM, Vlad Ureche <vlad.ureche@gmail.com> wrote:

On Fri, Feb 3, 2012 at 9:07 PM, Daniel Sobral <dcsobral@gmail.com> wrote:

On Fri, Feb 3, 2012 at 15:05, Vlad Ureche <vlad.ureche@gmail.com> wrote:
>
> Are you referring to the compiler API? The library API is available and
> up-to-date:
> http://www.scala-lang.org/archives/downloads/distrib/files/nightly/docs/library/index.html

That's 2.9.x's API, not master's.

You are right, the 2.9.2 jenkins job was replacing the trunk scaladoc pages.
Issued a pull request which should prevent any other version except trunk from publishing the scaladoc pages.

Thanks for reporting it!

Cheers,
Vlad

--
« Je déteste la montagne, ça cache le paysage »
Alphonse Allais

--
« Je déteste la montagne, ça cache le paysage »
Alphonse Allais

--
« Je déteste la montagne, ça cache le paysage »
Alphonse Allais

Sorry, I was too fast. I see there are two competing pull requests, none of which is merged yet.

aha, I see Lukas beat me to it with a cleaner patch (weird, I didn't see his pull request when submitting mine) 

Someone (other than Mr. Jenkins) has built and pushed to scala-tools, and that's what I saw earlier..

yes, that was me -- I can't merge pull requests, however, so we'll have to wait until it comes off the queue and hits master

On Wednesday, 8. February 2012 at 11:32, Adriaan Moors wrote:

Sorry, I was too fast. I see there are two competing pull requests, none of which is merged yet.

aha, I see Lukas beat me to it with a cleaner patch (weird, I didn't see his pull request when submitting mine) 

Someone (other than Mr. Jenkins) has built and pushed to scala-tools, and that's what I saw earlier..

yes, that was me -- I can't merge pull requests, however, so we'll have to wait until it comes off the queue and hits master

I think you should have commit rights, Adriaan - except if you disagree, it's true that itmeans one more thing to do.. It could also be Toni, or someone else.

For the record, I'm not against having commit rights, but I don't think it's urgent I have them.I'm hopeful the build automation will make situations like these increasingly rare.

On Wed, Feb 8, 2012 at 1:33 PM, Lukas Rytz <lukas.rytz@epfl.ch> wrote:

On Wednesday, 8. February 2012 at 11:32, Adriaan Moors wrote:

Sorry, I was too fast. I see there are two competing pull requests, none of which is merged yet.

aha, I see Lukas beat me to it with a cleaner patch (weird, I didn't see his pull request when submitting mine) 

Someone (other than Mr. Jenkins) has built and pushed to scala-tools, and that's what I saw earlier..

yes, that was me -- I can't merge pull requests, however, so we'll have to wait until it comes off the queue and hits master

I think you should have commit rights, Adriaan - except if you disagree, it's true that itmeans one more thing to do.. It could also be Toni, or someone else.

I made another pull request to remove the 'v' in the version number (v2.10.0-... becomes 2.10.0...). Unfortunately most (all?) tools assume that the first three components of a version are actual numbers, and Eclipse is one of them. So in order to  keep the beast happy, we need to sacrifice the first 'v' in our version string. 
For the record, our version strings before this change didn't have the 'v' either.
iulian

On Wed, Feb 8, 2012 at 11:32 AM, Adriaan Moors <adriaan.moors@epfl.ch> wrote:

Sorry, I was too fast. I see there are two competing pull requests, none of which is merged yet.

aha, I see Lukas beat me to it with a cleaner patch (weird, I didn't see his pull request when submitting mine) 

Someone (other than Mr. Jenkins) has built and pushed to scala-tools, and that's what I saw earlier..

yes, that was me -- I can't merge pull requests, however, so we'll have to wait until it comes off the queue and hits master

--
« Je déteste la montagne, ça cache le paysage »
Alphonse Allais

On Wed, Feb 8, 2012 at 6:36 AM, Adriaan Moors wrote:
> For the record, I'm not against having commit rights, but I don't think it's
> urgent I have them.

You have them anyway.

alias git-blame="echo adriaan.moors@epfl.ch"