Crash With Logging Pref. & Target Location is Out of Space

[Berylium]

I would be surprised if this hasn't been mentioned before but I did not see it listed in the TRAC nor did I see it doing a search for "space" in the Bugs forum.

The bug is very simple and doesn't really require a verbose log or many of the specifics jbrjake requests in the "READ ME before posting bugs" thread. Using the Mac GUI (on 10.5.7), if the preference "Put individual encode logs in same location as movie" is enabled and the destination for the video queued to be encoded is out of space Handbrake will crash. This behavior is exhibited up to build 2725 and at least as far back as build 2679 (though I imagine it goes back as far as the logging preference was available).

If the logging preference is not enabled Hanbrake will not crash, though, it will stop the encode immediately and say that the job is finished when obviously it is not. Perhaps Handbrake could alter it's "When Done" notification to notify the user that the Destination is out of space?

-Berylium

[dynaflash]

... and how would hb know when you are running out of space since it has no way of knowing (especially with cq encodes ) how much space is still to be used up at any given time ?

There has to be some responsibility to the user to do sane things.

[Da Man]

Dynaflash- Are you new here? "cough, cough".

[eddyg]

Graceful handling of write failures leading to a useful error and job termination are what are probably required.

Cheers Ed.

[Berylium]

Dynaflash,

I'm not suggesting HandBrake try to figure out if there's enough room for an encode. Obviously, that's impossible for the reason you give — there's no way to know what size the final encode will be. I'm merely suggesting that when the target has 0 bytes free that Handbrake make that announcement. Certainly, it's not HandBrake's job to police a user's disk usage but what eddyg said seems reasonable. At the very least, the crash should be avoided.

-Berylium

[dynaflash]

Fair enough. I frankly do not know the best way to handle this though. Bug noted.