Release Manager

A generic talk on the OpenMW project.
Locked
User avatar
lgromanowski
Site Admin
Posts: 1193
Joined: 05 Aug 2011, 22:21
Location: Wroclaw, Poland
Contact:

Release Manager

Post by lgromanowski »

Zini wrote: Looks like we have a problem. Our release manager stepped down (RL got in his way). I am still awaiting a statement if that is immediately or from the next version on, but for now we should assume the former. That means we need to find a new release manager quickly. The task does not require any coding skills, but good organisational skills and persistence.

My personal favourite would be sir_herrbatka, if he wishes to take the job.
raevol wrote: What does a release manager have to do? What skills are required?
Zini wrote: The release manager must keep track of the builders (and look for new ones if they go inactive). When a release is ready, I PM the release manager to start the release process.
The release manager then must collect the packages from the builders and upload them. Once this is done, he must post a release message (based on the change log; I help with the details) on our forum and all other forums and sites where we have a presence.

Edit: If you want the job, I would be okay with it too.
sir_herrbatka wrote: I would be glad to help since I lack more valuable, coding skills. If reavol wants to be RM IMHO He is the right man and I have nothing against but I would suggest myself as a deputy RM just in case of a nuclear war.

Silly circumstances shouldn't stop us from doing the job done. ;-)
raevol wrote: I'm down to do it, or sir_herrbatka could take it, I don't have strong feelings either way. I just want to make sure it doesn't require some technical skills I don't have? From your description it sounds like you need to be able to email, PM, download files, upload files, make forum posts, write changelogs, communicate in a timely and effective manner... anything I am missing?

sir_herrbatka may be a better candidate tbh, since he does the "what's up" announcement, and thus is more familiar with recent changes. I mean, I could always cheat and just read what he has posted (I read this forums two or three times a day anway), but that's one thing to consider. I'm more than willing to take the task if sir_herrbatka would rather not, either way. :)
sir_herrbatka wrote: raevol; you are on it, I can replace you in case of emergency situation. Everybody ok?
Zini wrote:
anything I am missing?
No. That is all.

Okay then. I guess you need to contact chewit about access to our google code site (the download location). And other sites too? I am not sure about that.

Our Mac build should be available soon (from corristo). I think Hircine was one of the other builders. I don't have any more details about the builders.

Here is the PM I sent chewit for 0.11.0:
You can base your release message on the automatically generated change log: http://bugs.openmw.org/projects/openmw/roadmap

Its a bit lengthy this time. Let me know if you need help with deciphering anything.

There are several entries regarding the launcher. Because the launcher was not part of the 0.10.0 release, you can simply fold them into a single statement, stating that we have a launcher now.

Other important things to mention:
- Even though the launcher handles plugins and multiple masters, OpenMW can't (yet)
- This release contains some performance optimisations, but we aren't done with it yet.

Since we had some blunders in the previous release descriptions, I offer to review any release message you want to post on our forum and elsewhere.
Edit: I think Ace was the Windows builder.
Zini wrote: Mac build already available. For now on corristo's github download page: https://github.com/corristo/openmw/downloads/
Ace (SWE) wrote: After some minor issues with Qt and two small launcher problem I've gotten the windows build done.
The launcher never set a windows codepage on the ESM reader so string reading crashed (Shouldn't this be a default value?) and the config file the launcher created called the Direct3D FSAA variable "Anti aliasing" which caused OpenMW to crash on launch.

Both of those errors are only applicable in windows builds though so I can see how they slipped past.

x86 Package
pvdk wrote:
Ace (SWE) wrote:...the config file the launcher created called the Direct3D FSAA variable "Anti aliasing" which caused OpenMW to crash on launch.
I encountered this problem when I built the launcher inside a virtual machine. The launcher did not crash but I did not have any anti-aliasing values to choose from. I suspected this was because of AA not being supported in VirtualBox.

Are you sure the value is called FSAA for Direct3D? I searched around for this when I wrote that code but if I recall correctly the value is called Anti aliasing for Direct3D and FSAA for OpenGL, since Full-screen Anti Aliasing only works with OpenGL. The config file generated by Ogre's native settings thingie should have the correct name. Remove Ogre's config file and launch openmw directly to make it appear. (The error you encounter could be a problem with Qt's config file reader, but it should crash on other keys as well then).

Many thanks for supplying a patch, but could you please clone my repository on github and push your changes to mine? That way incorporating changes is much easier (and the commits will bear your name, proper credits :)).
pvdk wrote: Forgot to mention: I have a source .tar.bz2 ready for download on my GitHub.
@raevol: please upload this to the Google Projects page with the other packages.
raevol wrote: Woosh, ok, I just got home and need to pass out, but will get on this tomorrow! I am excited!

EDIT: I PMed chewit, he's probably going to need a bit more info from me, but the ball is rolling!
raevol wrote: Alright, I have access to the google code page now. Should I upload the windows build linked earlier in this thread and the patch for the launcher? or should those be consolidated and distributed together?

EDIT: Beyond that it looks like I am missing binaries for linux, and that's it? I'll need to get the changelog ready too.
Ace (SWE) wrote:
raevol wrote:Alright, I have access to the google code page now. Should I upload the windows build linked earlier in this thread and the patch for the launcher? or should those be consolidated and distributed together?
The launcher patch was a simple code patch, disregard it.
raevol wrote:
Ace (SWE) wrote:The launcher patch was a simple code patch, disregard it.
Roger. :)
Zini wrote: @Ace: I see you added another fix today. Is it included in the binary too?

@raevol: No idea what is delaying the Linux package. You could get started with writing the release message anyway. From my previous experience I would guess that it will need an iteration or two.
raevol wrote: The launcher is new in this release, correct? Can all of the changes regarding it be consolidated into one entry of "Launcher added with xyz features"?
sir_herrbatka wrote: This is just the only new feature, not the only change.
raevol wrote:
sir_herrbatka wrote:This is just the only new feature, not the only change.
Right, I'm working from the changelog from the tracker that Zini linked. There's a lot of entries on there for the launcher, just wondering how to handle them. I PMed my draft to Zini.
Ace (SWE) wrote: @Zini: The link to the windows build has my fixes in it.

Which also reminds me that one of the fixes is not eligible to be added to pvdks branch as it is because of a change in the master branch, if it was to be added to pvdks newlauncher branch then it would only cause compilation to fail. This leads to me not having the faintest clue where I should be trying to get that fix merged.
Zini wrote: Okay. Problem. I just tried the launcher from my master branch and it isn't working either. Needs Ace's fixes probably. I am surprised that this problem does not show up on Mac though.

@Ace: Can you recreate your fix based on my master branch? pvdk's launcher branch isn't relevant anymore, because it has been integrated into master.
Ace (SWE) wrote: @Zini: The fixes I have made are based on the master branch, they should be just fine to merge.
Zini wrote: Confirmed and merged. Also moved the tag. Again. :( Next time I tag after the release process has been successfully completed.

I think at least the the source package needs updating. OS X too? Sorry!
raevol wrote: Ok, I'll wait for a new source and Mac package (if needed). Sorry for being a little out of the loop, but who does our linux builds?
corristo wrote: I'll rebuild OS X package :)
pvdk wrote: I'll supply a new source package then
raevol wrote: I love you guys. :)
pvdk wrote: Ok, a new source package is available on my GitHub.

@Zini: Weird that you were having problems too with the launcher, what platform?

Can someone (Ace?) confirm anti-aliasing works with Direct3D and the FSAA setting? Some people on the interwebs claim it gets ignored with D3D.
Zini wrote: @pvdk: Still the good old Ubuntu. Launcher crashed on startup (on the encoding stuff).

@raevol: We are still missing the Linux package, right? IIRC Hircine is our Linux builder.
Ace (SWE) wrote:
pvdk wrote:Can someone (Ace?) confirm anti-aliasing works with Direct3D and the FSAA setting? Some people on the interwebs claim it gets ignored with D3D.
With 0 FSAA
With "8 [Quality]" FSAA

It works.
Rhys wrote:
Ace (SWE) wrote:With "8 [Quality]" FSAA
Whats that white circle, which isn't in the other?
Ace (SWE) wrote:
Rhys wrote:Whats that white circle, which isn't in the other?
The sun.
pvdk wrote: Great! I'm convinced :)
raevol wrote:
Zini wrote:@raevol: We are still missing the Linux package, right? IIRC Hircine is our Linux builder.
Yes. Lemme see if I can rally him.
Hircine wrote: I will get to work on it tonight and upload it tomorrow.

worst time for me as my download speed was capped because i used the Internet too much. presently It would probably take me hours to upload the binaries. I am at my University, downloading the required files. Will come back tomorrow to upload. (should take a few short minutes)
Zini wrote: All stop again. There seems to be a problem with the launcher. I have two reports now that it crashes after a fresh install. Quoting myself from another thread:
I am deeply sorry (this goes especially to all the builders). It didn't show up in my tests. pvdk, could you have a look at it again? I think the same problem was already mentioned by lgro.
I think as OpenMW grows in complexity, we need to rework our release process. A bit more internal testing, before we are having a release? Well, we have plenty of time now to think about that. The next release is still a good while away.
raevol wrote:
Zini wrote:I think as OpenMW grows in complexity, we need to rework our release process. A bit more internal testing, before we are having a release? Well, we have plenty of time now to think about that. The next release is still a good while away.
Sounds like a plan.
Peppe wrote:
pvdk wrote:Ok, a new source package is available on my GitHub.
This is wrong.
As there have already been a openmw-0.11.0-source.tar.bz2 creating a new one is a bad, as this leads to us having two packages out there with the same name but different content. This makes for lots of fun trying to figure out what people are really using when you get bug reports...

If you repack after making something available to others, please increase the number (or add some other identifier if there is a procedure for that).
raevol wrote:
Hircine wrote:I will get to work on it tonight and upload it tomorrow.
Any luck with this? Or were you waiting on a patch?
Zini wrote:
If you repack after making something available to others, please increase the number (or add some other identifier if there is a procedure for that).
Well, we didn't really make it available to others. Our official download site is on google code. As I see it, the github download site is just a convinient way for the builder to send a package to the release manager.
Actually, since we have decided on a release manager now, it might be a better idea to just use email. This will avoid the potential problems Peppe mentioned.
Any luck with this? Or were you waiting on a patch?
We need to rewrite a decent part of the launcher code. Can't release in the broken state.
Locked