292614 - Break up GRE

Status: RESOLVED INVALID

(Core Graveyard :: Embedding: GRE Core, defect)

Description

[John R. Moser]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.6) Gecko/20050405 Firefox/1.0
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.6) Gecko/20050405 Firefox/1.0

As I understand it, Gecko is a major monolyth, with each piece integrated into
Gecko's codebase.  I've heard that SVG support is on its way, as a part of GRE
rather than a plug-in.

Why not make everything a plug-in if possible?  gstreamer, xmms, and abiword all
implement a lot of standard features as plug-ins.  Even the Linux kernel
implements drivers as loadable modules.  Plug-ins could be loaded and unloaded
as needed, saving some memory.

Using plug-ins for individual pieces of GRE would also allow the GRE core to be
handled better by the devs; the smaller code base would make it easier to design
a proper API so that Firefox, Thunderbird, and Mozilla wouldn't all need their
own copy of GRE, eating up precious memory, about 25M each.  It would also help
to get Mozilla onto embedded systems; if Thunderbird and Firefox both used the
same copy of GRE,they could both fit on DamnSmallLinux, instead of just Firefox.

Reproducible: Always

Steps to Reproduce:

Comment 1

[Benjamin Smedberg]

This bug is vague to the point of uselessness. The goal is to provide a
well-featured core and well-defined extensibility mechanisms... if you have
specific extensibility mechanisms in mind, please discuss on the
netscape.public.mozilla.embedding newsgroup and then file bugs as appropriate.

The plan for the 1.9 timeframe is to unify all the XUL apps under a single
runtime (libxul+xulrunner, descendents of the GRE), but realistically it is not
a major performance hit on modern systems to have multiple geckos in memory,
especially given the static-linking benefits.