We considered it, but it's not enough of a benefit to be worth
violating the simplicity & purity of blobs being "only data".
(especially as a blob could in theory be data and a schema blob, and
we'd then need to deal with any such bugs of them being classified or
named wrong)

As for reindexing speed, there's an upcoming change that'll scan all
the blobs in a more optimal order that should speed up reindexing by
2x or more, taking advantage of the blobpacked storage layout and also
optimistically assuming that small blobs are more than likely the
schema blobs, and reading them in smallest-first order.  (it still
needs to eventually read all blobs anyway to compute the "wholeref" of
files, but if we can read the big data chunks contiguously from
blobpacked, we can probably avoid the redundant reads of those chunks
we read when reindexing blobrefs in lexicographical order today)


On Mon, May 14, 2018 at 6:26 AM, alimoeeny <[email protected]> wrote:
> I have a silly question. Why the blob-id does not indicated if this is a
> meta-data blob or a content blob. Sorry if I am getting all the terminology
> wrong.
> I mean, unless I am missing something basic here, to create the index, the
> index server needs to actually fetch all the blobs and check if they are a
> json blob (meta data) or the are a content (binary?) blob.
> Right?
> Wouldn't it be helpful of the blob ids were like: SHA224-T-xxxxxxxx where
> that -T- would be -m- for meta data (json) blobs and -b- for content or
> binary blobs?
>
> Am I making sense?
>
> Ali
>
> --
> You received this message because you are subscribed to the Google Groups
> "Perkeep" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/d/optout.

-- 
You received this message because you are subscribed to the Google Groups 
"Perkeep" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to