673260 - HttpChannelChild::RecvAssociateApplicationCache needs to be buffered like other necko IPDL messages

Status: RESOLVED FIXED

(Core :: Networking: HTTP, defect)

Description

[Josh Matthews [:jdm]]

The body of RecvAssociateApplicationCache (http://hg.mozilla.org/mozilla-central/annotate/5d1799e16fe9/netwerk/protocol/http/HttpChannelChild.cpp#l196) needs to be moved into another function (AssociateApplicationCache) which is called from a runnable, exactly the same process as the surrounding Recv* functions in that file.

Comment 1

[Han Chang]

Attachment: Refactoring attempt #1

Refactored as described, compiled with success, ran the `make xpcshell-tests` in obj/netwerk/tests as well as `make check` in obj/netwerk/tests with all tests passing.

Not sure if additional tests need to be written / run though, please advise.

Comment 2

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Comment on attachment 554801 [details] [diff] [review]
Refactoring attempt #1

>+HttpChannelChild::RecvAssociateApplicationCache(const nsCString &groupID,
>+                                                const nsCString &clientID)
>+{
>+  if (mEventQ.ShouldEnqueue()) {
>+    mEventQ.Enqueue(new AssociateApplicationCacheEvent(this, groupID, clientID));
>+  }
>+  else {

Should be "} else {".

Comment 3

[Jason Duell]

Comment on attachment 554801 [details] [diff] [review]
Refactoring attempt #1

Good stuff.

I've fixed some nits and will check in a new patch.  Here's what I changed, FYI:


>+class AssociateApplicationCacheEvent : public ChannelEvent
>+{
>+  public:
>+    AssociateApplicationCacheEvent(HttpChannelChild* child,
>+                                   const nsCString &groupID,
>+								   const nsCString &clientID)

For some reason the above line uses tabs instead of spaces--fixed and make sure it indents with other parameters.

>+    : mChild(child)
>+    , groupID(groupID)
>+    , clientID(clientID) 

minor nit:  the clientID line has an extra space at the end of the line.

>+    {}
>+
>+    void Run()
>+    {
>+      mChild->AssociateApplicationCache(groupID, clientID);
>+    }

I've been formatting these as one liners (see elsewhere in file), which is non-Mozilla-standard (and I'm not sorry!).  Might as well keep file consistent.

>+void
>+HttpChannelChild::AssociateApplicationCache(const nsCString &groupID,
>+                                            const nsCString &clientID)
>+{

Note that we don't need an AutoEventEnqueuer in this method: since we don't call into client code that might spin the event loop, we're guaranteed that regular IPDL queuing will handle serialization.

> 
>-bool
>-HttpChannelChild::RecvAssociateApplicationCache(const nsCString &groupID,
>-                                                const nsCString &clientID)
>-{

Thanks for moving this--don't know how it got stuck here in between OnStartReqEvent and RecvOnStartRequest.

Comment 4

[Jason Duell]

Attachment: v2: minor fixups

Comment 5

[Jason Duell]

Han,

Thanks for the patch!  I've checked it into the "inbound" tree at Mozilla--if there are no errors detected by the test system, it will land in mozilla-central and make it into releases as of version 9.

  http://hg.mozilla.org/integration/mozilla-inbound/rev/c7c232adc3ec

Comment 6

[Han Chang]

Crap, sorry for all the formatting problems, my .vimrc settings aren't quite compatible with Mozilla's. Is there someplace I can grab a Mozilla approved .vimrc file? Or if such a place doesn't exist, where can I create one?

Comment 7

[Mounir Lamouri (:mounir)]

http://hg.mozilla.org/mozilla-central/rev/c7c232adc3ec

Comment 8

[Jason Duell]

Han: Mozilla files generally have a modeline at the top (sometimes they only have an emacs one: we sadly have more emacs users than vi-ones).

The file you modified has a modeline already:

   /* vim: set sw=2 ts=8 et tw=80 : */

Somehow you must have overridden 'et' (expandtabs).

Not a big deal!

Comment 9

[Han Chang]

Oh, that's super cool! Will read up more on modelines, should make things a lot easier. Thanks for the pointer!