SGI: Development

FF3, got three windows and about ten tabs per open, we're talkin Dodge Charger now !



Any idears for this ? Tons of them ...

As in "fast" or "breaks often"?

I was thinking about the ones with the long chassis, fringed cowboy tires, comfy seats and shapely grilles ... were they fast, too ?

This happens because void does not have a size, you can't go (void *)+1 because it won't know how much to increase it by. With gcc, it defaults to char bits IIRC. Not sure how to fix that, perhaps add a char * variable to do the arithmetic and then pass it to data.


You can ignore these, it's those gcc people who think that enum = int

Thanks, mr. duck. I'll try to follow up on this ....


And they were such nice young lads before they turned to fascism

You can switch on enum compare warnings in GCC, and you can switch them off for MIPSpro. It's just the default that's different.

If the warnings bother you and you don't feel like fixing the code I suggest the latter.


Yeah, that's what K&R refer to as 'implementation defined behavior'. Wise men don't go there

The error is more bothersome than the warnings, that's for sure


I don't want to either but the newer FontForge has support for pango, cairo, infinality and the rest so that now, with FF3 in the crib, some of that is more worthwhile.

btw, are any of youse guys gonna vote for FF3 to go to /current ? I know it sometimes crashes still but it's not worse than FF2 and at least the crashes are better described.


is a lot more descriptive than just disappearing like the Cheshire cat. The pages it does work on (non-teeny-bop pages seem to be fine) are nice.

bad_alloc? Seriously? The only way bad_alloc should ever be thrown, at least at this point in modernity, is if every scrap of RAM and swap on your workstation are in use, and I'm guessing that never happens......

So just to clarify vishnu, that would be a "no" vote on moving FF3 to /current?

Anyone else seen a bad_alloc crash with the ff3 beta?

I've used the neko_ff3 beta pretty extensively and so far haven't seen it crash. When hamei first reported bad_alloc crashes I tried and wasn't able to replicate the crashes hamei was experiencing with ff3 .

This'd definately fall into the wid-assed-hunch category, but in the ff3 beta thread I noticed hamei mentioned: If no one else steps up to mention they've experienced a bad_alloc crash with ff3, and that rolled-forward installation goes back as far as your first Fuel ( 2005? ), might it be possible you have just the just the right combination of historic installs, uninstalls, upgrades and hardware switches to be causing the bad_alloc grief?

I'd be the last one to keep ff3 out of current because of the bad_alloc crashes* ! In fact, I want it to go GA.

I don't think it's the installs uninstalls and so on causing the cashes tho. Firefox 3 isn't very good. You just don't see it because you have fast internet connections, reliable dns and straight-thru links to advertising. Run it on a fresh install on a high-resolution display, split down the middle (right where you are displaying the application) into channel 0 and channel 1, with a slow connection and lying dns and many sites that never appear - many sites that the buffoons of today's web insist on linking to -- and you'll see some bad behavior as well And oh yeah, all the crap javascript that litters the landscape .... FF3 is a piece of shit. It's just the best piece of shit we're going to get and I'm personally happy to have it.

(Back to fontforge, god only knows what ff3 does with fonts. Any time you change anything in your font path, the entire browser changes. It's insane.)

* Those are actually the minority. Normally it's just a memory failure, kaboom disappear. Sometimes when sitting idle, doing nothing.

** knock on wood, but I haven't across that one either. Maybe the mozdev guys weren't predisposed to letting bygones be bygones and coded in a special hamei-only easter egg/bomb.

Just checked the Mozilla cross reference to see where the exception handling routine for bad_alloc is and came up with nuthin. Is it just me or has the Mozilla codebase become darn-near indecipherable at this point in its history? And they wonder why the rest of the world picks webkit over Geko...

They aren't that talented

Back to Fontforge, this seems to be relevant to the problem :

http://forums.devshed.com/c-programming ... 33526.html

these are interesting :

http://fontforge.org/ff-history.html

http://ospublish.constantvzw.org/typo/i ... in-my-mind

Seems that there are quite a few real improvements in the more recent versions of fontforge <hint hint>

Actually no; to quote jwz from the Code Rush documentary, "we're nowhere near that organized."

jwz also said the Mozilla people were a bunch of flakes and he was happy as hell to get out of there and get an honest, worthwhile job, running a bar

You'd think he could have been a bit more charitable given the vast fortune he made in the five years he worked at Netscape...

He was being charitable !

With good reason; it took them three more years to get version 1.0 kicked out the door ...

They should have kicked it into the garbage can instead

Netscape 3 was good. Netcrap 4 was worse than awful. Mozilla was even worse shit, which is why Phoenix came to life.

Then they ruined that.

Those people should be drowned like unwanted kittens.


Now, about the fontforge void* ... no suggestions ? The newer version sounds like a definite improvement. But it's not written in APT, so I'm pretty useless at getting it fixed