How to index arbitrary headers?

[Dmitrijs Ledkovs]

On 3 October 2012 19:32, Petri Savolainen <petri at koodaamo.fi> wrote:
> Hi,
>
> thanks for your response. I am evaluating notmuch / xapian for building an
> application for analyzing in various ways a fairly large number of emails
> accumulated over several years. I am afraid the number of headers that would
> ultimately need to be indexed is therefore quite a lot larger than what
> notmuch currently indexes.
>
>  Petri
>
> 2012/10/1 Austin Clements <amdragon at mit.edu>
>>
>> Quoth Petri Savolainen on Oct 01 at  3:39 pm:
>> >    Hello,
>> >    I could not find information anywhere in notmuch docs about what is
>> >    actually indexed - specifically, what email headers are indexed and
>> >    searchable? If a header is not indexed, does searching for its value
>> > still
>> >    result in a search hit?
>> >    It would be nice if one could just provide the list of headers to be
>> >    indexed in some configuration file or something.
>> >    Thanks,
>> >     Petri
>>
>> notmuch doesn't currently implement this, though it is an
>> oft-requested feature.  One (not insurmountable) difficulty is that
>> the database would have to be rebuilt if a user-configured list of
>> headers changed and there are technical limitations that prevent us
>> from simply indexing all headers.  Out of curiosity, what headers are
>> you interested in indexing?
>>
>> The currently indexed headers are described in man
>> notmuch-search-terms.
>

Use mapreduce instead: hadoop or discoproject or haddop with dumbo
should be faster.

Regards,

Dmitrijs.