Search service returns a digest per wavelet instead of a digest per wave

[GoogleCodeExporter]

Currently, if a search matches multiple wavelets inside a single wave, the 
search service will generate multiple digests.

In this case, it should only generate one digest.

This will not be a problem until we have implemented private replies in WIAB.

Original issue reported on code.google.com by jose...@gmail.com on 23 Oct 2010 at 11:53

[GoogleCodeExporter]

Why is it bad? I think it should generate digest per each blip that matches the 
search criteria, i.e. for query "about:something" - if several blips contain 
the text - it should return digest per each blip.

Original comment by vega113 on 25 Dec 2010 at 6:12

[GoogleCodeExporter]

A digest entry is something that shows up in the left-hand search panel. At the 
moment digests are simple, just a title and (not displayed) participant 
information. Google Wave also has message and unread counts.

A wave is the thing that users can most easily comprehend as a thing they're 
searching for. The search results digests should appear like inbox digests and 
behave much the same way. If a wave matched multiple times, in different blips, 
it would add a lot of clutter to the results to have an entry per matching blip 
rather than one per wave.

There is definitely room to add more information to the digests though. For 
example we could include information about which blips matched the search, and 
later jump to and/or highlight them in the wave panel. In this way a search 
result is based on a waveref, but it still makes sense to group the waverefs 
all referring to the same wave into one digest entry.

Original comment by ano...@google.com on 29 Dec 2010 at 2:36

[GoogleCodeExporter]

Re-assigning to Tad because he's begun work on search internals which will 
eventually lead to resolving this issue.

Original comment by ano...@google.com on 29 Dec 2010 at 2:37

[GoogleCodeExporter]

I think that the content of the digest should be directly linked to the content 
of the search. 

If the user is searching for text, it is most helpful for them to see all the 
blips that contain that text (or a wave, with a matching blip count). When they 
open that wave, the blips should be highlighted.

If the user is searching for a participant, then Wavelets are most helpful. 

If the user is searching for a tag, then Waves are more helpful, as tags are 
stored in their own data-document wavelet within a wave (AFAIK). 

For compound searches, the *most* specific of these should be used. So 
"with:joe@bloggs.com" would display Wavelets, but "with:joe@bloggs.com red 
herrings" would display blips (or wavelets with blips highlighted). 

While these are just some ideas as to how, it is important that you all 
understand that if you want to provide a really functional & helpful wave 
search you need more specific ways of outputting data.

Original comment by nat.abbo...@gmail.com on 29 Dec 2010 at 12:45