Closed Bug 353983 Opened 18 years ago Closed 18 years ago

--enable-application=netwerk (standalone necko)

Categories

(Core :: Networking, defect)

defect
Not set
normal

Tracking

()

RESOLVED FIXED
mozilla1.9alpha1

People

(Reporter: Biesinger, Assigned: Biesinger)

References

Details

(Keywords: dev-doc-complete)

Attachments

(1 file, 2 obsolete files)

To make it possible to build standalone Necko, I want to add an --enable-application=netwerk compile option. This builds just the libraries that Necko needs (essentially XPCOM, libpref, intl and necko itself).

This adds a tier_necko which contains parts of tier_gecko.

NOTE: I had to name this netwerk rather than necko because the build system expects a mozilla/$(MOZ_BUILD_APP)/build.mk file. I could add a necko alias if desired.

Also, it is not really possible to run the unit tests here, because almost all of them need xpcshell and therefore xpconnect, which is not build in this configuration. Perhaps --enable-tests should build it.
Attached patch patch (obsolete) — Splinter Review
Attachment #239812 - Flags: superreview?(darin)
Attachment #239812 - Flags: review?(benjamin)
Status: NEW → ASSIGNED
Target Milestone: --- → mozilla1.9alpha
Attached patch patch v2 (obsolete) — Splinter Review
Now with client.mk changes. Also, fixes the condition for the libreg build, this is the change:
-ifneq (,$(MOZ_NO_XPCOM_OBSOLETE)$(MOZ_XPINSTALL))
+ifneq (1_,$(MOZ_NO_XPCOM_OBSOLETE)_$(MOZ_XPINSTALL))

That should lead to the desired result of libreg being built if xpcom-obsolete is built or if xpinstall is built (or both)
Attachment #239812 - Attachment is obsolete: true
Attachment #239849 - Flags: superreview?(darin)
Attachment #239849 - Flags: review?(benjamin)
Attachment #239812 - Flags: superreview?(darin)
Attachment #239812 - Flags: review?(benjamin)
Attached patch patch v3Splinter Review
oops. fix BOOTSTRAP variables. sorry for the bugspam...
Attachment #239849 - Attachment is obsolete: true
Attachment #239850 - Flags: superreview?(darin)
Attachment #239850 - Flags: review?(benjamin)
Attachment #239849 - Flags: superreview?(darin)
Attachment #239849 - Flags: review?(benjamin)
Comment on attachment 239850 [details] [diff] [review]
patch v3

+  mozilla/extensions/auth                       \

looks like this has to be removed, or normal checkouts will be broken :(
Comment on attachment 239850 [details] [diff] [review]
patch v3

I'm fine with this provided bsmedberg is happy with the build details.
Attachment #239850 - Flags: superreview?(darin) → superreview+
Attachment #239850 - Flags: review?(benjamin) → review+
checked in
Status: ASSIGNED → RESOLVED
Closed: 18 years ago
Resolution: --- → FIXED
Depends on: 354835
There's no documentation for this, right? Adding dev-doc-needed, as at the very least various outdated docs on MDC should be updated to say that standalone necko builds are possible.
Keywords: dev-doc-needed
And so it transpires that Google decided to code their own networking implementation from scratch for use in its upcoming Chrome 2.0 browser.

Great work, Mozilla.org! (NOT!). You guys missed a great chance to have the Mozilla networking libs used by another major software player.

I guess nobody even thought of approaching Google to make them use Necko when they were having trouble porting Chrome to non-Windows platforms due to its reliance on WinHTTP, right?

Well, too late now...

FC
You know that Darin Fisher (one of the lead engineers on Chrome) used to be the owner of our network module, right? I don't think this was due to lack of knowledge of the existence of necko.
So what would a .mozconfig file for this look like?  Just

ac_add_options --enable-application=netwerk

Or would more stuff be needed as well?  I'm working on documenting this at:

https://developer.mozilla.org/En/Building_Necko_standalone
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: