842887 - update Mozilla's copy of libevent from 1.4.7 to 2.0.21

Status: RESOLVED FIXED

(Core :: General, defect)

Description

[Josh Aas]

We want to use libevent for our new DNS resolver, and in order to do that we need libevent2. This offers much better performance for Windows, and improvements for other platforms as well.

Comment 1

[Josh Aas]

Attachment: fix v1.0

This is a big patch, around 5mb. It's mostly just removing the old copy of libevent, adding the new copy, and outside of that one directory it just tweaks a makefile. It includes clear documentation in "README.mozilla".

https://tbpl.mozilla.org/?tree=Try&rev=a35185c8daf4

Comment 2

[Josh Aas]

Here is a basic overview of the changes from 1.4.x to 2.0.x:

https://raw.github.com/libevent/libevent/patches-2.0/whatsnew-2.0.txt

Comment 3

[Wouter Wijngaards]

Attachment: patch with libevent different (part of 1.9 set)

this makes libevent: export headers.  Actually compile itself.  Compiles libevent on windows.
This patch is incomplete (on windows at least).

Comment 4

[Josh Aas]

Lets get libevent compiling on Windows in a separate bug.

Comment 5

[Josh Aas]

Any idea when you'll be able to do this review, Chris?

Comment 6

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Should be this week, but feel free to ping someone else in the meantime.

Comment 7

[Benjamin Smedberg]

Comment on attachment 717145 [details] [diff] [review]
patch with libevent different (part of 1.9 set)

I'm a bit leery of exporting the libevent headers. Is there a particular reason we need to export them instead of continuing to use them (via LOCAL_INCLUDES or a global setting) directly from their source-tree location?

Comment 8

[Josh Aas]

(In reply to Benjamin Smedberg  [:bsmedberg] from comment #7)

> I'm a bit leery of exporting the libevent headers. Is there a particular
> reason we need to export them instead of continuing to use them (via
> LOCAL_INCLUDES or a global setting) directly from their source-tree location?

That patch is future work, the patch with review requested is the one covered by this bug and it doesn't change how we export headers - it simply updates libevent. We can handle the latter patch that changes header export later, in another bug.

Comment 9

[Josh Aas]

https://tbpl.mozilla.org/?tree=Try&rev=41f69e0ec0ed

Comment 10

[Josh Aas]

pushed to mozilla-inbound

https://hg.mozilla.org/integration/mozilla-inbound/rev/71545c41ea4c

Comment 11

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/71545c41ea4c

Comment 12

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Is ipc/chromium still the best location for this code?

Comment 13

[Josh Aas]

(In reply to :Ms2ger from comment #12)
> Is ipc/chromium still the best location for this code?

Probably better off in a top-level dir once non-ipc code makes use of it, but we can do that in another bug.