Fx38.0.5 and Fx38 ESR Version Numbers

[mcdavis]

I don't understand how to write chrome.manifest appversion range checks to distinguish Fx38.0.5 from later versions of Fx38 ESR.

What I'd like to do is something like this:

Code: Select all

skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx31/
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx32/ appversion>31.99
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx33/ appversion>32.99
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx34/ appversion>33.99
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx35/ appversion>34.99
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx36/ appversion>35.99
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx37/ appversion>36.99
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx38/ appversion>37.99
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx38.0.5/ appversion>=38.0.5
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx39/ appversion>38.99

Normal release channel Fx38 would get its code, Fx38 ESR would also get that same code, and Fx38.0.5 would get different code.

However, when I look at version numbers used with Fx31 ESR (<== that's THIRTY ONE, using that to guess future 38 ESR version numbers) I see these used versions:

http://ftp.mozilla.org/pub/mozilla.org/ ... /releases/

[DIR] 31.0esr/ 18-Jul-2014 03:27 -
[DIR] 31.1.0esr/ 26-Aug-2014 00:43 -
[DIR] 31.1.1esr/ 24-Sep-2014 07:57 -
[DIR] 31.2.0esr/ 11-Oct-2014 20:48 -
[DIR] 31.3.0esr/ 25-Nov-2014 18:43 -
[DIR] 31.4.0esr/ 06-Jan-2015 11:21 -
[DIR] 31.5.0esr/ 25-Feb-2015 18:58 -
[DIR] 31.5.1esr/ 20-Mar-2015 09:57 -
[DIR] 31.5.2esr/ 20-Mar-2015 20:54 -
[DIR] 31.5.3esr/ 21-Mar-2015 09:38 -
[DIR] 31.6.0esr/ 26-Mar-2015 14:40 -
[DIR] 31.7.0esr/ 05-May-2015 09:13 -

If the same scheme is used with 38 ESR then we get for example Fx38.1.0 ESR, which hypothetical version number would be greater than 38.0.5 and break the range comparison in chrome.manifest.

This wasn't an issue the last time Firefox did a halfway-between release (for the change to Yahoo as default search, whatever that version was), because the major version number of that one wasn't the same as the major version number of an ESR. Here, the major version number is the same for both the halfway release and the ESR.

Am I correct that this is an issue?

Any thoughts on what to do about it?

[WaltS48]

I don't understand how to write chrome.manifest appversion range checks to distinguish Fx38.0.5 from later versions of Fx38 ESR.

Any thoughts on what to do about it?

Don't worry about 38.0.5 since it will be replaced by 39.0 in 3 4 weeks when I presume 38.1.0 is also due for release.

[patrickjdempsey]

You just need a second line for 38 to handle the ESR versions I think:

Code: Select all

skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx38/ appversion>=38.0
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx38.0.5/ appversion>=38.0.5
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx38/ appversion>=38.1
skin    nasanightlaunch-appver  nasanightlaunch jar:chrome/inner.jar!/appver/fx39/ appversion>=39.0

[patrickjdempsey]

Don't worry about 38.0.5 since it will be replaced by 39.0 in 3 4 weeks when I presume 38.1.0 is also due for release.

Since this version introduces some new features but won't be identical to 39, it means Theme Authors will likely have to support it slightly differently (unless they want breakage on the new features).

[mcdavis]

Thanks Patrick, I went with your solution.

Although I see here something that looks to be 38.0.1ESR:

http://ftp.mozilla.org/pub/mozilla.org/ ... andidates/

which puts me back to worrying that we'll have ESR version numbers straddling 38.0.5. I guess I'll have to ask what the official intentions are.

[patrickjdempsey]

I would guess that Mozilla is gearing up for a chemspill release to cover Logjam... there's also been a slew of crashers for this version so they might be planning something for that as well. Anyway, this system allows for 4 chemspills between 38.0 and 39.0 (38.1 ESR) before there would be a conflict between the 38.0 branch and Northern Spring 38.0.5. And it's very unlikely that 4 chemspills will happen.

[patrickjdempsey]

This is what the tracks would look like imagining a worst-case scenario of 4 chemspills per release, and assuming that 38.0.5 will be released at the same time as the first chemspill:

Code: Select all

38.0    38.0esr-QA
38.0.1  38.0.1esr-QA  38.0.5
38.0.2  38.0.2esr-QA  38.0.6
38.0.3  38.0.3esr-QA  38.0.7
38.0.4  38.0.4esr-QA  38.0.8
        38.1esr-QA            39.0   
        38.1.1esr-QA          39.0.1    
        38.1.2esr-QA          39.0.2   
        38.1.3esr-QA          39.0.3      
        38.1.4esr-QA          39.0.4   
        38.2esr                       40.0    

Remember that ESR doesn't "push" until after 12 weeks of QA.