Re: [Tracker] Question about the internals of Tracker and extractors

[Martyn Russell <martyn lanedo com>]

On 07/10/10 13:23, Ken Schmedding wrote:
> Greetings,

Hi,

> I have been working on extending tracker support for several things my
> org needs. We needed support for UTF-16 in the text extractor and we
> needed an rtf extractor.  This has brought up two questions from issues
> I am seeing running with prototypes of the new extractors.
>
> 1.  While processing large text files I can  get a shutdown timeout. Is
> there a way to increase this and is it #define QUIT_TIMEOUT 30 /* 1/2
> minutes worth of seconds */ in tracker-main.c.  Is there also a relation
> to #define MAX_EXTRACT_TIME 10 in tracker-extract.c?

Yes, that's what you would change to change the shutdown. Alternatively, 
we also added the -d or --debug command line which ignores the timeout 
and only shuts down on ctrl+c.

> 2.  I am seeing some errors while processing large files in the debug
> log.They look like this:
>
> Tracker:   Got DBus timeout error on 'file:///some file', but it could
> not be caused by it. Retrying file.
> Tracker: Could not process 'file:///some file': Message did not receive
> a reply (timeout by message bus)
>
> It seems that I am processing so much I may have exceeded some timeout.
> Is there any way to change this, and where would it be?

Yes. We see this occasionally with some extractors like the PDF 
extractor and some rather large PDFs. We try to protect ourselves 
against dodgy extractors by having an allowed time to extract data. How 
long is your extraction taking for that particular file?

--
Regards,
Martyn