slow pdf conversions + processes not cancelling

[cybmole]

I found Naomi Klein's shock doctine in pdf. I already own the paper book but I fancied having in on Kindle, with headers & footers stripped so I try a conversion.

I figured I'd start with pdf to rtf so that I could inspect results in word.

after 5 mins , - job details was still still only at 1% so I cancel hob & try pdf to epub instead, figurign I'll check output with calibre internal viewer
after 5 more mins that job is also only at 1% so I cancel it.

( is it nromal for progress bar to stick like that ? )

later, I wonder by my PC has slowed to a crawl & open task manager, I see 2 pdf convert processes STILL RUNNING at 40+ % cpu each, thus both of my cores are being maxed out by jobs that I'd cancelled - how come ?
I kill them in task manager & PC returned to normal.

[DoctorOhh]

Originally Posted by cybmole

after 5 mins , - job details was still still only at 1% so I cancel hob & try pdf to epub instead, figurign I'll check output with calibre internal viewer after 5 more mins that job is also only at 1% so I cancel it.

There is no provision to "cancel" a job. You have to select the job and click Stop selected job or just click on Stop all non-device jobs. Since you say the job was still running later I'm guessing you just bailed on viewing the job details. I just tested stopping the selected job and it stopped the job directly.

Originally Posted by cybmole

( is it nromal for progress bar to stick like that ? )

Yes. Also expecting PDF conversions to end in 5 minutes usually is too optimistic.

Originally Posted by cybmole

I wonder by my PC has slowed to a crawl & open task manager, I see 2 pdf convert processes STILL RUNNING at 40+ % cpu each, thus both of my cores are being maxed out by jobs that I'd cancelled - how come ?

I would guess that the jobs were still running in calibre because you didn't stop the jobs or, under certain situations that I couldn't duplicate, the stopping jobs function is broke.

If you wish to ensure worker processes don't overwhelm your system try going to Preferences - Behavior and changing Job priority from Normal to Low.

[Manichean]

Just as a reference, I've had PDF conversions run a couple of hours and ZIP-HTML conversions run over one day.

[banjomike]

Originally Posted by Manichean

Just as a reference, I've had PDF conversions run a couple of hours and ZIP-HTML conversions run over one day.

Sounds like paradise!

Look at this screen cap.

After over 4 hours, reached 1%, the temp folder contains 182,811 files and is 2.1 Gig. I gave up.

Incidently, the pdf is 9MB.

[Manichean]

Originally Posted by banjomike

Sounds like paradise!

Look at this screen cap.

After over 4 hours, reached 1%, the temp folder contains 182,811 files and is 2.1 Gig. I gave up.

Incidently, the pdf is 9MB.

What's your point? You cancelled the conversion and thus have no converted book. I let it finish and had the book converted.

[DoctorOhh]

Originally Posted by banjomike

After over 4 hours, reached 1%, the temp folder contains 182,811 files and is 2.1 Gig. I gave up.

I agree with Manichean. If you wanted the book converted you should have left it running. The fact that it said 1% is meaningless, don't use that number as any kind of guide of how long it is going to take. Manichean stated he had one conversion take a whole day. If the conversion doesn't stop with an error it is most likely still chugging along.

I had 1 epub to epub conversion take over 7 hours and it probably said 1% the majority of the time it was running.

[cybmole]

Originally Posted by dwanthny

There is no provision to "cancel" a job. You have to select the job and click Stop selected job or just click on Stop all non-device jobs. .

that is exactly what I did , twice, in both cases the status then jumped to 100%, inside of job details. & calibre then reported no jobs active, so I closed it.

maybe the conversion process can only stop at certain "waypoints" & thus just kept going.

I've since read, elsewhere, that this specific book (7Mb PDF) does not convert well in calibre or via other routes.

I'm not fussed about further debugging but it was calibre v7.19, I think< running in 32bit Vista. I have since updated it to 7.22. the process names both began with pdfto..

[DoctorOhh]

Originally Posted by cybmole

that is exactly what I did , twice, in both cases the status then jumped to 100%, inside of job details. & calibre then reported no jobs active, so I closed it.

Sounds like you did it correctly. To paraphrase what I stated above, just because I couldn't duplicate it doesn't mean it ain't broke.

Originally Posted by cybmole

I'm not fussed about further debugging but it was calibre v7.19, I think< running in 32bit Vista. I have since updated it to 7.22. the process names both began with pdfto..

You still might want to consider changing Job priority from Normal to Low so you can function during long conversions.

[cybmole]

Originally Posted by Manichean

Just as a reference, I've had PDF conversions run a couple of hours and ZIP-HTML conversions run over one day.

auto kindle did the Naomi Klein Shock Doctrine entire 7MB pdf in about 7 seconds ( but managed not to output any text beyond the TOC ! ) .


I guess speed AND quality is still big ask :-)..

meanwhile if anyone can hook me up with a non pdf source of shock doctrine please do so ...

[speakingtohe]

I have had calibre process continue to run long after calibre was exited. Mostly only noticed them when installing new version.
When I was using laptop though I always rebooted after running a lot of calibre conversions but no big deal

@cybermole
With problem pdfs that won't convert I have used adobe to save a copy and the copy AFAIR converted normally.

Helen

[cybmole]

well after failing to convert my specific book I googled pdf to epub & found this limited free offer:
http://www.pdftoepub.com/authorspromotion.asp
i registered as an author & got a key + download link - testing now...

oh well - that could not do my specific book either, though it worked on some other test stuff. it has a neat solution to header/footer issues - rather than grapple with regex you just drag sample image borders to crop the area from which you want text to be converted i.e. you simply exclude the header part of the page to eliminate it. That's intuitive & easy to do.

PS you get 5 free tries then it prompts for the activation key.