1267102 - The way we calculate search strings for jobs is redundant and inefficient

Status: RESOLVED FIXED

(Tree Management :: Treeherder, defect)

Description

[William Lachance (:wlach)]

1. We duplicate the logic to create a search string repeatedly in clonejobs.js, despite having a centralized space for creating it (and a logical place to call that function, in the job model).
2. We store the search string permanently, even though we only need it when filtering jobs. It would be much more efficient to dynamically calculate it when required for that, then throw it away later.

Some quick back of the envelope calculations suggest that we're wasting upwards of 2 megabytes on this in the default treeherder view (assuming 2000 jobs and 205 characters average for the string:

```
>>> len('OS X 10.10 debug W3C Web Platform Tests e10s W3C Web Platform Tests e10s W-e10s(1) Rev7 MacOSX Yosemite 10.10.5 mozilla-inbound debug test web-platform-tests-e10s-1 eb84c6ab9e0bc1385031cdcad93df1e42bd00b8e')
205
>>> 205*        10000
2050000
```

Comment 1

[GitHub Autolander Bot]

Attachment: [treeherder] wlach:1267102 > mozilla:master

Comment 2

[William Lachance (:wlach)]

Comment on attachment 8744751 [details] [review]
[treeherder] wlach:1267102 > mozilla:master

Hey Cam, I did some quick memory profiling this weekend and this seemed like an easy win.

Comment 3

[Cameron Dawson [:camd]]

Comment on attachment 8744751 [details] [review]
[treeherder] wlach:1267102 > mozilla:master

Nice!  This seems like a great change, and makes the code more read-able since it adds the functions to the job model, where they belong.  :)

Comment 4

[GitHub Autolander Bot]

Pull request has landed in master: https://github.com/mozilla/treeherder/commit/ae122751bb9af51a20e443ca159569b07edd6aed

Comment 5

[Treeherder GitHub Bugbot]

Commit pushed to master at https://github.com/mozilla/treeherder

https://github.com/mozilla/treeherder/commit/e08d52c47404b49653c8eaaeadd2514b0e8a2d0c
Bug 1267102 - Be more efficient about storing job title + search information

Since we only need it when actually filtering the job list, just dynamically
calculate there instead of keeping a (potentially large amount) of redundant
data in memory constantly. It doesn't take long to recalculate this relative
to other expenses (such as recalculating the DOM).

Since we were storing about 200 characters per job, and that a typical push
has upwards of 2000 jobs, this can save up to 2 megabytes on the default
treeherder view of 5 pushes.

Comment 6

[Treeherder GitHub Bugbot]

Commit pushed to master at https://github.com/mozilla/treeherder

https://github.com/mozilla/treeherder/commit/76849cd222394a60b3ea63fb973afd5a82ef5c9a
Bug 1267102 - Fix platform filtering