0.6.37 convert to ePub bug(?)

[cutterjohn42]

SEEMS TO BE RESOLVED as I was able to successfully convert one of the problematic files this morning...







Today I attempted to convert 3 HTML files to ePub by merely adding them to calibre, editing metadat, then selecting those files and then hitting convert, OK to default conversion "features".

The last two files(books) into the stack converted in seconds, while the last one hung, so I let it go several minutes with no progress.

Cancelled(stopped) job.

Added that file again by itself(defaults, ePub), goes to 67% progress, show job details brings up an empty dialog(prior successful converts have data in the dialog)

Let go some minutes, cancel.

Have a brain fart

Add TWO books, some other down the list manually and then the one that I really wanted converted.

The file that I want converted is, while the one at the top of the stack hangs(pretends likes it's processing, but nothing happens)

Ubuntu 9.10 x86-64, Calibre 0.6.37, Sony PRS-505, all HTML -> ePub
hanging item is in the 1 (one) position in the job detail list.

(6m my last workaround try the #1 item is now at 6m 30s. I call this one a bug. Sorry no time to see if the bug ticket system got fixed or whatever, but will try a ticket later if I remember to. See my possible quoting bug from Author names IIRC from several weeks ago post as to my ticket system problem at that time.)

[EDIT]
incorrect Ubuntu version corrected.

Also viewing the converted file with the builtin viewer, ePub format results in invalid argument 22.

So, I tried converting another 2 random HTMLs that had previously been converted. Second one, #2 finished, #1 showed 100% but calibre still seemed to be working and the job detail dialog was empty.

The converted file when attempted to view with the provided viewer, ePub format resulted in invalid argument 22 again, where before it had been fine.

I need to check some additional random ePubs to determine if it's a viewer or converter problem.

Trying 10 or so other random ePubs converted with OLDER versions of caliber, came up fine in the viewer, so I'm now thinking that there is a converter problem either alone or in combination with the queueing bug...
[/EDIT]

[EDIT2]
I'll try them on the 505 and see what happens... (I hate reboots... but I'm dreading one now...)

Yep, file was never converted properly or at all as the 505 merely displayed invalid page(or was it image?)
[/EDIT2]

[EDIT3]
Poking through the ticket system, I'm beginning to believe that this may be a dupe of other "NEW" HTML conversion bugs. Or maybe not as they all seem to be quite old... the ticket filtering system seems to be missing some filtering options... but I'll need to check under firefox...
[/EDIT3]

[cutterjohn42]

Just doing a new post to bump this:
This must be a conversion bug rather than anything to do with the queueing as I downloaded some recipes yesterday and they all ran to completion fine, however I did not verify that the output was correct, i.e. that I ended up with a readable ePub.

I'm sticking with the HTML -> ePub conversion problem as previously convertable HTML no longer are, but if I have time I'll try to see about running in debug mode and get a copy of the logs.

[EDIT]
Did a quick check of the recipe download outputs and they seemed fine, not that that tells me anything new as Calibre is still converting most HTML -> ePub fine. Probably some regression in Calibre's HTML parsing for conversion behavior.
[/EDIT]

[EDIT2]
Hmmm... I just took the time to re-run one of the files that was failing the other day(repeatably) and now it converted fine.
[/EDIT2]