703042 - Desktop Linux/Gtk build with --disable-pango is broken

Status: NEW

(Core :: Graphics, defect)

Description

[Jonathan Kew [:jfkthame]]

Attachment: WIP - fix up the --disable-pango build on desktop Linux/Gtk

This was broken due to gfxFT2Font* changes and refactoring in bug 684889. Although the --disable-pango configuration is not something we actually ship, or care much about, it may sometimes be useful for debugging purposes as it produces a desktop Firefox build whose font-rendering codepaths are much more similar to the Mobile product.

The attached patch allows the browser to build and run, although it's rather rough-and-ready. A better fix would involve substantial reworking of gfxPlatformGtk to properly use gfxFT2FontList, but it's not a high priority IMO.

Comment 1

[Oleg Romashin (:romaxa)]

you have disabled whole FT2 stuff, why? does it display any fonts in that mode?
IIRC qt build was compiling fine without these changes...

Comment 2

[Raghavendra Prabhu]

I can confirm this. The build fails here -- http://sprunge.us/EIhW 

The mozconfig used is here -- http://sprunge.us/BjYA

Comment 3

[Jonathan Kew [:jfkthame]]

Attachment: patch, fix the Linux/GTK build with --disable-pango so that it compiles again

With this patch, the browser seems to build and run OK, except that user fonts (@font-face rules) are not supported.

Comment 4

[Oleg Romashin (:romaxa)]

Comment on attachment 626042 [details] [diff] [review]
patch, fix the Linux/GTK build with --disable-pango so that it compiles again

Works fine for me

Comment 5

[Oleg Romashin (:romaxa)]

Attachment: Qt platform Ft2Fonts hookup, tested, working

Comment 6

[Oleg Romashin (:romaxa)]

Attachment: Make it not crash on attempt to access platform font list

Comment 8

[Oleg Romashin (:romaxa)]

Attachment: Restore old F2Fonts and name it as FT2Fc fonts

Comment 10

[George Wright (:gw280) (needinfo me!)]

Is this likely to land soon? It's blocking my refactoring in bug 762737 as we use the non-Pango gfxFont implementation for Skia/Linux.

Comment 11

[Nicholas Nethercote [inactive]]

The first patch has bit-rotted quite badly, unfortunately.  (I wanted it because I'm writing a memory reporter for Freetype because it uses a substantial amount of memory on B2G, and I wanted to test the reporter code on desktop first.)

Comment 12

[Jonathan Kew [:jfkthame]]

Clearing old review requests. Bug 947379 is removing the MOZ_PANGO conditional altogether. If there's anything here we still want to do, let's start afresh; otherwise, I think this is WontFix.