Cross-posting to mozilla.dev.tech.network.   The main discussion is on
mozilla.dev.tech.dom.

---
So, right now rumor has it that necko is a standalone code component,
that can be used w/o firefox (do we have any actual clients using it
this way?).  I'm guessing we want to keep it that way if possible.

But I'm wondering how much our electrolysis (e10s) work is
inadvertently entangling necko into relying on the dom and ipdl (and
chromium!) codebases, and if we want to take some time to think about
what if anything we can do to keep necko independent, or pave the road
towards it being independent again at some point, or whether it's just
going to be too difficult to be worth it.

So far my work (none of which is checked in yet) has the following
dependencies:

- netwerk/Makefile.in now includes chromium-config.mk
- necko code now #includes files generated by ipdl
- I'm about to add calls to XRE_GetProcessType(), which will rely on
code that's planned to be in nsXULAppAPI.h.

This is all enough to cause necko to now be dependent on chromium and
libdom.   I also suspect that it will be a significant pain in the
neck to try to support a version of necko going forward that doesn't
have these dependencies--though some could possibly be minimized, and
I suppose we could in theory (via #ifdefs) continue supporting a
single-process necko that is standalone (but it would definitely be
extra work, and maybe a lot of it).   More realistic perhaps would be
a multi-process version of necko that relies on libchromium and ipdl,
but not libdom or XULRunner (but this would require moving
ContentProcess.ipdl and friends somewhere besides the dom tree).

Anyone have any thoughts on this issue?
_______________________________________________
dev-tech-network mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-tech-network

Reply via email to