Closed djspiewak closed 9 years ago
@djspiewak cool. can we put a list what needs to be done before release? I just created milestone, perhaps we can add just issues to it?
Update: moved release notes draft to a wiki page.
I can't edit your message, but #298 should also be mentioned under Bug Fixes:
time.duration
to obtain a new reference time on every Task
evaluationI can't edit your message, but #298 should also be mentioned
Updated. Also, should I perhaps stick the release notes in a Gist or a wiki page? I didn't realize that only collaborators could edit messages.
The release notes of past releases are in the scalaz-stream wiki, so its reasonable to also put them there.
The merge of master
into master-a
isn't clean. Are there any objections to just creating a new master-a
based off master
and then force-pushing master-a
?
Ok, looks like we're good to go for 0.7! Attention, @pchiusano. Your party would like to meet you at the front desk.
master-a
is now updated!
Sweet. Question - do we want to try to do the release on sonatype? I can also just do a bintray release for now and we can use the same tag / series to do a release on sonatype, in case there is any red tape with that.
A while ago, @fthomas asked me about putting together a CONTRIBUTORS file. I'd like to do that as part of this release. I don't really want to be the one in charge of deciding how to describe everyone's contributions, but was thinking: names listed alphabetically by last name, and each contributor maintains their own one-line description of what they've worked on.
@pchiusano are you still planning to rename the project? From what @djspiewak wrote in https://github.com/scalaz/scalaz-stream/issues/258#issuecomment-77590139 it seems to be easier to rename the project before releasing to Maven Central.
@fthomas Yeah, though not for 0.7. Also, we still need to come up with a good replacement name. :)
Also, I am currently on childcare duty so I prob won't be able to cut release until Monday at earliest. :)
@pchiusano great! We would use a public release as soon as it's available (we're already working off of locally-published jars :) ) so if a bintray release is easier, I'd say we should just go for that first. If/when there's demand for a sonatype release, then that could come second.
I have no preference for the presence of a CONTRIBUTORS file or not, but if there is one, I'd say it makes sense for each developer to just submit a PR adding themself and a tagline to the file.
Created here: https://github.com/scalaz/scalaz-stream/blob/master/CONTRIBUTORS.markdown
No need to do PRs if you already have commit access. In fact, if you are feeling really lazy, you can just edit the markdown directly from the GitHub site. :)
On Fri, Mar 20, 2015 at 12:36 PM Alissa Pajer notifications@github.com wrote:
@pchiusano https://github.com/pchiusano great! We would use a public release as soon as it's available (we're already working off of locally-published jars :) ) so if a bintray release is easier, I'd say we should just go for that first. If/when there's demand for a sonatype release, then that could come second.
I have no preference for the presence of a CONTRIBUTORS file or not, but if there is one, I'd say it makes sense for each developer to just submit a PR adding themself and a tagline to the file.
— Reply to this email directly or view it on GitHub https://github.com/scalaz/scalaz-stream/issues/331#issuecomment-84063739 .
Okay, I did the release on bintrary, I'd like to get the release notes cleaned up before announcing, though. Please hold off on any announcements for the time being.
Awesome :)
Okay, @djspiewak @alissapajer @pchlupacek @fthomas I cleaned up the release notes page a bit. Could you all take a look through and see if there is anything missing?
https://github.com/scalaz/scalaz-stream/wiki/0.7
Also, @djspiewak can you do a better job of characterizing the issues with tcp
? And if there are issues, can we get some failing tests added? If you don't have time to do that, I'd maybe just say: "The new tcp
module should still be considered alpha quality. We've noticed a number of issues with it that we are still tracking down. If you'd like to help make the module production quality, please get in touch."
Also, @djspiewak on my edits, I tried to succinctly characterize the situations in which the finalization semantics changes is relevant to users (basically resource allocation plus use of nondeterministic functions). That way the person reading can decide to what extent they need to really dig in (like scodec-stream's usage of scalaz-stream doesn't care about any of that). I also moved that long technical discussion to the end, so people can quickly scan the rest of the items.
@pchiusano On the issues with tcp
, I did file a long issue which contains (as I recall?) three sub issues. I would break things down into two significant categories:
tcp.server
taking a concurrency bound and then still returning a nested process is the best example of this, especially if an application wants to do something useful with the values produced by the connection handlers.Suspend
blocks used to be run nonconcurrently). You can take a look at the changes I made just to get the bug/step
branch running. In a sense, you can actually treat the existing tests as "failing tests", since I directly disabled functionality in the tests in order to make them work.Anyway, the issue I filed has a lot more details. I'm fine with just saying "should be considered alpha quality". I just wanted to get a warning in the release notes, since people might be looking at using it in its current state, which I don't believe would be the best idea. I do plan on working on the tcp
stuff in this next release cycle. It's a little lower on my priority list than optimized chunk processing and onComplete
soundness, but it is on my list.
On a completely separate note, I'm pretty sure that the close0
hang in the NIO2 support on Mac is actually a bug in Darwin, based on the effect it has on the process image and the way that the JDK implements that function. So that sort of sucks, because it means our odds of fixing it are pretty low.
@pchiusano Just read through the release notes and added a couple things. Looks good to me.
I added the Process1Syntax.contramap
and the typeclass instances for Process1. LGTM
@pchiusano Are we green for tweeting/etc about 0.7?
Standby, I will send an email to the scalaz list which you can link to if you like.
On Tue, Mar 24, 2015 at 1:36 PM Daniel Spiewak notifications@github.com wrote:
@pchiusano https://github.com/pchiusano Are we green for tweeting/etc about 0.7?
— Reply to this email directly or view it on GitHub https://github.com/scalaz/scalaz-stream/issues/331#issuecomment-85612119 .
Okay, fire away: https://groups.google.com/forum/#!topic/scalaz/sKMmvVChK08
On Tue, Mar 24, 2015 at 3:14 PM Paul Chiusano paul.chiusano@gmail.com wrote:
Standby, I will send an email to the scalaz list which you can link to if you like.
On Tue, Mar 24, 2015 at 1:36 PM Daniel Spiewak notifications@github.com wrote:
@pchiusano https://github.com/pchiusano Are we green for tweeting/etc about 0.7?
— Reply to this email directly or view it on GitHub https://github.com/scalaz/scalaz-stream/issues/331#issuecomment-85612119 .
0.7/0.7a is released and we're all shouting about it. I'm going to close this issue.
Thanks so much for cutting the release, Paul! I know I'm super-psyched for the changes.
Also, we still need to come up with a good replacement name.
Can we do something about this? Like opening an issue where people can post their suggestions? Or anything else? I'd love to see scalaz-stream getting a new name sooner rather than later.
I'd like to request that we start the process to release version 0.7 once we have a resolution to #330 vs #318. There are several other important fixes which have landed since 0.6, and it would be nice to get them out. The only other thing I can think of that I want to get into this release is a minor version bump in our scodec/scodec dependency and the appropriate change to
io.toInputStream
. Basically, @mpilquist exposed a function inByteVector
that will make the inner buffering much faster in certain cases. I'll have that taken care of and PRed this weekend (update: PR'd in #332).