Re: [Treebase-devel] Some results of testing with a user

[William P. <wil...@ya...>]

On Mar 19, 2010, at 4:00 PM, Vladimir Gapeyev wrote:

> I see Bill has already run a fix on the 2264 TLSs.  Should we do it for 22-TLSs as well?  

I just did that. Unfortunately, it doesn't quite solve the problem. 

If I search on S2377 and then click on the Taxa tab, I get nothing listed.
If I search on S2377 and then click on Matrices and then click on Taxa, I get a long list.
If I search on S2377 and then click on Tree and then click on Taxa, I get a long list. 

With this query:

select taxonlabelset_id, title, study_id from taxonlabelset_taxonlabel join taxonlabelset using (taxonlabelset_id) where taxonlabel_id = 72150;

I see that not all records with the same taxonlabel_id got 

 taxonlabelset_id |         title          | study_id 
------------------+------------------------+----------
             3105 | M3562                  |     2377
             8626 | Untitled Block of Taxa |       22
(2 rows)

I see that not all taxonlabelsets got changed from id 22 to whatever they should be -- the taxonlabelsets created by trees blocks did not get changed. 

This is for a misbehaving taxonlabel. Suppose I compare this with a behaving one from an earlier migration:

select taxonlabelset_id, title, study_id from taxonlabelset_taxonlabel join taxonlabelset using (taxonlabelset_id) where taxonlabel_id = 95894;                                                                                                                                                          

taxonlabelset_id |         title          | study_id 
------------------+------------------------+----------
             3841 | M713                   |       22
             7509 | Untitled Block of Taxa |       22
            10400 | TaxonLabelSet10400     |      654
(3 rows)

... there's an extra row, with the two older ones still mapping to the dud study_id 22.  This new row looks like it was created by TreeBASE2, probably for the explicit purpose of having both trees and matrices map to the same taxonlabelset. Both matrix and treeblock tables have records with taxonlabelset_id = 10400, which has the correct study_id of 654. That seems to be missing for some of our more recently migrated data. 

This looks complicated to fix, but it may only affect a few studies. If so, perhaps we could just delete and re-enter these. 

I guess I'll look to see how many are affected in this way.

bp