julianh2o / Podbase

Image database aimed at scientific applications
MIT License
2 stars 0 forks source link

Priority 1: "Search" malfunctions #84

Closed danhartline closed 11 years ago

danhartline commented 11 years ago

In the ShrimpMyelin project, I entered a search term ("nucleus" -- I know the term is there as a tag, for example

/ShrimpMyelin/TEM/Saron/SM1/DH13-01_A2/FM111130-1_SM_DH13-01_A2_003-006_2k-mont.jpg'>FM111130-1_SM_DH13-01_A2_003-006_2k-mont.jpg

). It took a LONG time before it displayed a list -- turns out that was because the list was HUGE. The images in the list (or the ones I checked, at least) had no metadata associated (yet) so they should not have appeared. I tried a different term ("mitochondrion") with similar results: the list returned seemed to include ALL images for that project. It even returned the names of folders containing the images.

julianh2o commented 11 years ago

I'll take a look into this. I've been having some real trouble getting the search indexing to work. It seems that somehow my Lucene index is not being searched correctly. Last time I checked, it looked like it had been indexed correctly (I managed to get that fixed when you reported the last search bug). For now, try to avoid using search and rest assured that hopefully when I'm done this search mechanism should be extremely fast, powerful, and flexible!

danhartline commented 11 years ago

Sounds good -- Thanks! However, lack of a search capability will keep us from using PB2 for analysis (we are not yet to that point, of course -- we have to enter lots and lots of accumulated metadata first, and that will require a "stable" version).

D

I'll take a look into this. I've been having some real trouble getting the search indexing to work. It seems that somehow my Lucene index is not being searched correctly. Last time I checked, it looked like it had been indexed correctly (I managed to get that fixed when you reported the last search bug). For now, try to avoid using search and rest assured that hopefully when I'm done this search mechanism should be extremely fast, powerful, and flexible!


Reply to this email directly or view it on GitHub: https://github.com/julianh2o/Podbase/issues/84#issuecomment-21655322

Daniel K. Hartline Research Professor and Director Bekesy Laboratory of Neurobiology University of Hawaii at Manoa Honolulu, HI 96822 www.pbrc.hawaii.edu/~danh

julianh2o commented 11 years ago

While looking into this issue, I need to also take a look at the performance of inserting elements into the index.

julianh2o commented 11 years ago

I'm fairly sure this works now.. There will be some delay from the time that you enter information to the time that it becomes available for searching.. so don't test it be entering text and then immediately searching for it. Let me know if you notice any more problems.

I'll post a bug to improve search behavior a bit more. There are some performance and responsiveness issues to weigh.