Not beating around the bush with this. [wanted to have a catchy title… not being good with topics and all that, but.. no one cares about the title. It’s the content..]
Well, the release will be done within this week, just not today (31st August 2015, PST(or whatever it is now)). Really sorry about that. I realize we said end of August, but … ok. No buts.
Ok. There is a but.
But we are in the final stretch. And of course, the disclaimer here is that ‘final’ is loosely termed as within this week (and dare I even say, ‘at the earliest’ – reason? .. oh you don’t want to know the excuses.. Would gremlins be satisfact… no? Um.. the dog? umm the misalignment of… ).
The real reason is that with our infrastructure code, we don’t have the necessary infrastructure to support a testing environment. So any changes we do to the infra code is live every time we post(and even though the infra code is reviewed, like everything else), there could be unforeseen circumstances that give our infra a fit. (Yes. I know. In this day and age of DevOps and CI, this shouldn’t be the case; but, really.. the resources issue. I’ve been planning to find a way to get some sort of CI/testing env for our infra code locally.. just haven’t been able to… work and family and other RL issues.)
So, please do bear with us just for a little bit longer.
PS: I realize this isn’t a proper PR ditty. I’m not a PR person and therefore the aforementioned would probably get someone fired somewhere. Anyway… no offense to anyone. All I hope is that I can help this project in any way I can, even if it’s small.
The Win32 2.35 release build is done. By ‘done’, I mean the compile and pushing to the staging area has been completed. The actual release *isn’t* complete.
The win32 2.35 l10n repacks are also done. (Again, like above, in staging area).
Next up, I’m waiting for the busted l10n repacks to be done (linux/osx: ca, ja, fi, sv-SE and pt-PT). Once they’re done, we’ll be working on the next part (which will involve a bit of infra structure code changes).
Another update on 2.35.
1) build3 was ‘somewhat’ done but needs to be rebuilt due to some misconfiguration on my part. It doesn’t mean that build3 is a bust. It isn’t. It’s just that it’s uploading the files to the wrong places. (First time it uploaded was to the nightlies tree. :P)
2) l10n repacks needs fixing. I’ve been told it won’t require a build4(read: Callek’s gonna do his cset magic), but I guess we shall see.
3) Once the repacks are done, the *real* work is on getting the CVS-based-system-now-migrated-to-Mercurial part of the release process working. But thankfully, this part is automated, so I won’t be the reason for the slow down.
Again, thanks for your patience while I work on this.
NB: Due to my concentrating on getting the release done, I haven’t done a single Windows nightly in a while, and even then, it was in fit and starts due to tree bustage. I humbly apologize for that (for the not getting a windows nightly released… not the tree bustage. )
NB2: Comm-central AND comm-aurora are in a painful state due to a bunch of interrelated bugs, so there haven’t been any builds or nightlies in some time. While the patches are up for review(iirc), it just needs time to review them and I could do a post-land-review, but these patches *also* affect the Release builds and I think it was a good idea NOT to fix those bugs (at the sacrifice of c-c and c-a). But once this release is done, we *will* fix the tree.
Wanna know what?
I goofed. I messed up… a little. ok. A bit more than a little; but nonetheless, I pushed the wrong patch (can I squirm for an excuse? like… gremlins? aliens abducted my brain temporarily…though.. I don’t think they’d find anything great… ;P ) . This in turn, well… botched the build process. Well, not the build itself, but the upload process.
Ergo… I now introduce to you all, build 3. I can tell you that this is the HIGHEST build number I’ve ever done before (and probably in the history of SeaMonkey… *sigh*).
As I write this, the Linux build was green and the upload was ok. Now, it’s at the Linux l10n repacks. Once those go green (here’s hoping), it will be my turn to work on the Win32 builds/repacks. (Didn’t I say this was going to be ‘fun’ and ‘challenging’ earlier?… yup… I did… and boy is this nerve wracking.)
So again, we really do appreciate your patience, especially with my fumblings. But, as I quote, ‘if you don’t succeed, try try try again…’
Here’s an update to the progress…
We’re doing a build 2. What? What happened to build 1?
Build 1 went so much south, that we require a build2.
So with build2, it should go as far as building the code and uploading it to the staging server; but then we come across the l10n repacks (*gulp*). I’m hoping it’ll go ok; but since our builds were busted due to changes in the mozilla-esr38 code, I’m concerned there might have been other changes that affected the l10n repacks. And this is all with the automated stuff… The trepidation is there when it comes to manually generating the builds/repacks. Oh, and then the latter part of the release process… the updates. (*gulp*) and updates verifications which require some infra code changes (due to the migration of our CVS repos (patcher, patcher-configs, etc..) to Mercurial. Oh what fun… [very tempted to sing Jingle Bells here…]
But as Gollum once said… “We shall see… we shall see….”
Here’s an update to the progress.
As was expected, the release is being hindered by compounded bustages (infrastructure code wise). I haven’t seen what’s busting the actual build, but at the get go, the first step was busted. That has been fixed by Callek, and I’m currently looking at the bustage of our second step (source upload).
Like I said earlier… (or somewhere at least), this is going to be so fun, both in the sarcastic sense and the normal sense as it’s a challenge (well, at least for me) that I have accepted. 🙂
[I’d insert the “Challenge Accepted” meme here… but I don’t know the policies of doing so, so I’m just opting for a “Challenge Accepted Meme” phrase.]
The moment everyone’s waiting for.
The spinning of SeaMonkey 2.35.
It is a time of trepidation… for having not spun a release since 2.33.1, and with the requirement of a lot of changes AND plus the fact that I will need to do the Win32 builds/repacks manually… you can bet your bottom dollar this is going to be “interesting”.
So while we have started the spinning, we aren’t in the clear yet. When you have that brand spanking new 2.35 installed on your systems, that’s when we shall celebrate. Until then… keeping my fingers crossed that it will be not as painful as I expect. (Wanted to insert the “Nobody expects the Spanish Inquisition”, but realized it didn’t make sense… since I’d have to say “I didn’t expect the Spanish Inquisition…” of which it would make sense to say “Nobody expects the Spanish Inquisition! Our chief weapon is surprise…surprise and fear…..etc…”)
I *really* can’t wait til I can say that I’ll bring you to your regularly scheduled program… (get it? scheduled? regular… ie. release…err.. nvm.)
Unfortunately for the latest SeaMonkey 2.35 version with several security patches no language packs were available. Because my experience is that the 2.35 language packs Adrian Kalla contributed some weeks ago (and what are no longer available at his download area) work fine with that latest 2.35 build I asked him to send those language builds to me, and I made them available for public download at sourceforge.net.
On that page Adrian Kalla contributes localized installers.exe and languagepacks.xpi in various languages.
Only for testing, you will use those builds on your own risk.
After much confusion (still finding my way around), I *think* the repacks are ‘working’ (for some loose definition since I don’t know what ‘working’ is considering I don’t know what it was before it went pear-shape).
From what I see, the ‘failed’ locales are the ones with only one entry (the checksum entry), and I’m currently working out the ‘kinks’ in the process.