08:37 am, 9 Jan 09
chromium linux status
At work we have a program that renders pages. We use it for testing: e.g., do our font metrics match IE's? (It's unfortunately important for site compatibility.)
It's not useful as a browser, as none of the UI is there (e.g. you must type well-formed URLs into the URL bar). It is useful for testing v8, the networking stack, the graphics layer, etc. Note in particular that form controls obey my theme, but that we turn off antialiasing and form control metrics are still bit wacky (respectively, to match the way Windows looks and lays out for expected test output).
There's also been a lot of work done on the next pieces (the UI, the multiprocess bits), but the nature of the problem is that you need to get a large amount of code compiling before anything visible works. And some of those internals are gnarly -- what's the Linux analog of condition variables that work both within and cross-process and that can also be select()ed on? -- but if it were easy it'd be a boring job. You can see some of the IPC tests passing on our builders.
It's not useful as a browser, as none of the UI is there (e.g. you must type well-formed URLs into the URL bar). It is useful for testing v8, the networking stack, the graphics layer, etc. Note in particular that form controls obey my theme, but that we turn off antialiasing and form control metrics are still bit wacky (respectively, to match the way Windows looks and lays out for expected test output).
There's also been a lot of work done on the next pieces (the UI, the multiprocess bits), but the nature of the problem is that you need to get a large amount of code compiling before anything visible works. And some of those internals are gnarly -- what's the Linux analog of condition variables that work both within and cross-process and that can also be select()ed on? -- but if it were easy it'd be a boring job. You can see some of the IPC tests passing on our builders.
Pipes. No, I'm serious.
Implementing condition variable broadcast (however MS spells it) may be a little fiddly however.
I assume the matching font metrics thing requires matching fonts too? I ask just because some of my systems don't have Microsoft's fonts installed, and I've noticed the site compatibility problems you state, especially since sans-serif ends up meaning Bitstream Vera Sans which is short and stubby and not like Helvetica/Arial at all. I really just keep it that way so I can laugh at sites that assume the text will be Arial-shaped. :)