Closed Bug 1277714 Opened 8 years ago Closed 8 years ago

Upgrade hg.mozilla.org to Mercurial 3.9

Categories

(Developer Services :: Mercurial: hg.mozilla.org, defect)

Product:
Developer Services
Component:
Mercurial: hg.mozilla.org
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

(Not tracked)

Status:
RESOLVED FIXED

People

(Reporter: gps, Assigned: gps)

References

Details

Attachments

(9 files, 1 obsolete file)

global: upgrade Mercurial to 3.8.4 (bug 1277714);
58 bytes, text/x-review-board-request
smacleod
: review+
Details
mozhg: fix test output for Mercurial 3.8;
58 bytes, text/x-review-board-request
smacleod
: review+
Details
hgmo: update automationrelevance for Mercurial 3.8;
58 bytes, text/x-review-board-request
smacleod
: review+
Details
hgserver: update test output to match Mercurial 3.8;
58 bytes, text/x-review-board-request
smacleod
: review+
Details
hgserver: update test output to reflect Mercurial 3.8;
58 bytes, text/x-review-board-request
smacleod
: review+
Details
hgtemplates: synchronize templates with Mercurial 3.8.4;
58 bytes, text/x-review-board-request
smacleod
: review+
Details
global: upgrade to Mercurial 3.9.1 (bug 1277714);
58 bytes, text/x-review-board-request
Details
hgtemplates: synchronize templates with Mercurial 3.9;
58 bytes, text/x-review-board-request
smacleod
: review+
Details
global: bump extension testedwith versions (bug 1277714);
58 bytes, text/x-review-board-request
smacleod
: review+
Details 
It has been 1 month since Mercurial 3.8 was released. Per our upgrade policy, we're clear to upgrade to 3.8 (3.8.3 is the current latest available version).
Attachment #8760918 - Attachment is obsolete: true
Blocks: 450645
We never upgraded hg.mozilla.org to 3.8. 3.9.1 comes out in a few days. So let's just go to that.
Summary: Upgrade hg.mozilla.org to Mercurial 3.8 → Upgrade hg.mozilla.org to Mercurial 3.9
Depends on: 1298555
Depends on: 1298944
Comment on attachment 8786573 [details]
global: upgrade Mercurial to 3.8.4 (bug 1277714);

https://reviewboard.mozilla.org/r/75496/#review73976
Attachment #8786573 - Flags: review?(smacleod) → review+
Comment on attachment 8786574 [details]
mozhg: fix test output for Mercurial 3.8;

https://reviewboard.mozilla.org/r/75498/#review73978
Attachment #8786574 - Flags: review?(smacleod) → review+
Comment on attachment 8786575 [details]
hgmo: update automationrelevance for Mercurial 3.8;

https://reviewboard.mozilla.org/r/75500/#review73980

::: hgext/hgmo/tests/test-automationrelevant.t:137
(Diff revision 2)
> -        "parent": [],
> +        "parents": [
> +          "55482a6fb4b1881fa8f746fd52cf6f096bb21c89"
> +        ],

Do we know what's consuming this? if something is making a direct index into `parent` without checking for existence it's going to break.

I'm going to open an issue to make sure you consider this, but feel free to drop if you're confident it won't break anyone.
Attachment #8786575 - Flags: review?(smacleod) → review+
Comment on attachment 8786576 [details]
hgserver: update test output to match Mercurial 3.8;

https://reviewboard.mozilla.org/r/75502/#review73982

::: hgserver/tests/test-push-basic.t:59
(Diff revision 2)
>    remote: recorded changegroup in replication log in \d+\.\d+s (re)
>  
>  Blackbox logging recorded appropriate entries
>  
>    $ hgmo exec hgssh cat /repo/hg/mozilla/mozilla-central/.hg/blackbox.log
> -  * user1@example.com (*)> serve --stdio (glob)
> +  * user1@example.com @0000000000000000000000000000000000000000 (*)> serve --stdio (glob)

Errr, wth are these additions (out of curiosity, but laziness to not go digging)? It's unlike you for it not to be in the commit message ;), but maybe it should be obvious to me?
Attachment #8786576 - Flags: review?(smacleod) → review+
Comment on attachment 8786577 [details]
hgserver: update test output to reflect Mercurial 3.8;

https://reviewboard.mozilla.org/r/75504/#review73984

::: hgserver/tests/test-bookmarks-replication.t
(Diff revision 2)
> -  "bookmark": "bm1",
> -  "node": "04da6c25817b564b37238ee5144e5adf2af0cb5b",
> -  "date": [0.0, 0]
> -  }, {
>    "bookmark": "bm2",
>    "node": "e7d8e0aefcf6bcc137a21978e9a431c5b0dafd86",
>    "date": [0.0, 0]
> +  }, {
> +  "bookmark": "bm1",
> +  "node": "04da6c25817b564b37238ee5144e5adf2af0cb5b",
> +  "date": [0.0, 0]

huh, what's the sorting for these? Is this going to be stable between machines?
Attachment #8786577 - Flags: review?(smacleod) → review+
Comment on attachment 8786578 [details]
hgtemplates: synchronize templates with Mercurial 3.8.4;

https://reviewboard.mozilla.org/r/75506/#review73986

rubberstamp
Attachment #8786578 - Flags: review?(smacleod) → review+
Comment on attachment 8786579 [details]
global: upgrade to Mercurial 3.9.1 (bug 1277714);

https://reviewboard.mozilla.org/r/75508/#review73988

might be worth mentioning the 3.4 and 3.5 drop in the commit message
Attachment #8786579 - Flags: review+
Comment on attachment 8786580 [details]
hgtemplates: synchronize templates with Mercurial 3.9;

https://reviewboard.mozilla.org/r/75510/#review74000

::: hgext/hgmo/tests/test-json-filelog.t:99
(Diff revision 2)
> -              "pushid": 1
> +              "tags": [], 
> +              "user": "test"

This change of output seems like it could esily be a breaking change. `author` has become `user`, `pushid` is missing, the file you requested is no longer echod in the output...

Are we sure this isn't going to break anyone? What does the usage look like for this resource? I'll leave the judgement call up to you, so feel free to drop this.
Attachment #8786580 - Flags: review?(smacleod) → review+
Comment on attachment 8786581 [details]
global: bump extension testedwith versions (bug 1277714);

https://reviewboard.mozilla.org/r/75512/#review74004

::: hgext/reviewboard/server.py:59
(Diff revision 2)
> -testedwith = '3.3 3.4 3.5 3.6'
> +testedwith = '3.6 3.7 3.8 3.9'
> +minimumhgversion = '3.9'

Is there really any point testing with older versions if we're just requiring the latest?

::: hgext/robustcheckout/__init__.py:33
(Diff revision 2)
> -testedwith = '3.5 3.6 3.7 3.8'
> +testedwith = '3.6 3.7 3.8 3.9'
>  minimumhgversion = '3.7'

same, why test with 3.6
Attachment #8786581 - Flags: review?(smacleod) → review+
Comment on attachment 8786575 [details]
hgmo: update automationrelevance for Mercurial 3.8;

https://reviewboard.mozilla.org/r/75500/#review73980

> Do we know what's consuming this? if something is making a direct index into `parent` without checking for existence it's going to break.
> 
> I'm going to open an issue to make sure you consider this, but feel free to drop if you're confident it won't break anyone.

The API is used by TaskCluster. They don't look in the `parent` key. So I'm dropping this issue.
Comment on attachment 8786577 [details]
hgserver: update test output to reflect Mercurial 3.8;

https://reviewboard.mozilla.org/r/75504/#review73984

> huh, what's the sorting for these? Is this going to be stable between machines?

This appears to be fallout from https://www.mercurial-scm.org/repo/hg-all/rev/ca0c0ca30c62. So output should be stable.
Comment on attachment 8786580 [details]
hgtemplates: synchronize templates with Mercurial 3.9;

https://reviewboard.mozilla.org/r/75510/#review74000

> This change of output seems like it could esily be a breaking change. `author` has become `user`, `pushid` is missing, the file you requested is no longer echod in the output...
> 
> Are we sure this isn't going to break anyone? What does the usage look like for this resource? I'll leave the judgement call up to you, so feel free to drop this.

We implemented this template before upstream (bug 548629). So I'm fine with `author` -> `user` as a breaking change. I'll send a new version to restore `pushid`. Good catch.
Comment on attachment 8786581 [details]
global: bump extension testedwith versions (bug 1277714);

https://reviewboard.mozilla.org/r/75512/#review74004

> Is there really any point testing with older versions if we're just requiring the latest?

For the server, no. Although there may be something wonky with run-tests's compatible version detection code since both these extensions are in the same directory. I'll investigate.

> same, why test with 3.6

minimumhgversion doesn't actually work until 3.7 :)

I think during a previous code review someone told me to set this to a value that actually had an impact. It was either you or glob.
Pushed by gszorc@mozilla.com:
https://hg.mozilla.org/hgcustom/version-control-tools/rev/89749b0917f9
global: upgrade Mercurial to 3.8.4 ; r=smacleod
https://hg.mozilla.org/hgcustom/version-control-tools/rev/1f35d91c1dfe
global: upgrade to Mercurial 3.9.1 ; r=smacleod
https://hg.mozilla.org/hgcustom/version-control-tools/rev/af7287700528
global: bump extension testedwith versions ; r=smacleod
Status: ASSIGNED → RESOLVED
Closed: 8 years ago
Resolution: --- → FIXED
Deployment to production complete. AFAICT there was no perceived downtime for any client. The server logs appear clean of any new errors. Looks like a pretty turnkey deployment.
Blocks: 1300061
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: