| Time |
Nick |
Message |
| 07:48 |
|
BDorsey joined #evergreen |
| 09:26 |
|
Dyrcona joined #evergreen |
| 09:28 |
|
dguarrac joined #evergreen |
| 10:11 |
|
Dyrcona joined #evergreen |
| 11:10 |
Dyrcona |
Hm. Why am I getting NULL data ('\N') in output for a query that explicitly says IS NOT NULL on the column? |
| 11:28 |
Dyrcona |
Umm... That's not good. I've got a copy of production that's missing the asset.opac_visible_copies table. |
| 11:29 |
Dyrcona |
It's at least one other asset table. |
| 11:30 |
Dyrcona |
missing that is... |
| 11:34 |
Dyrcona |
Wait a minute... |
| 11:36 |
Dyrcona |
Oh, all right. asset.opac_visible_copies was dropped in an upgrade. I missed that release note. |
| 11:36 |
Dyrcona |
This copy of production has been upgraded. |
| 11:42 |
Dyrcona |
So, that's a script I'll have to change before we upgrade to 3.10. |
| 11:42 |
* Dyrcona |
attempts a sort-fields on a 807MB file. Go, Emacs, go! |
| 11:55 |
Dyrcona |
Well, before I can do that, I have to replace spaces with _ because the fields are delimited by tabs, and Emacs delimits fields by any whitespace character. Looks like it may have stalled doing the search and replace. |
| 11:56 |
Dyrcona |
Ah, no, it was till going. The screen just stopped refreshing at one point. Not sure if that's on Emacs or the window system. |
| 11:58 |
Dyrcona |
OIC, I should leave it alone and let it finish. It freezes when I try to do something else in a different file buffer. |
| 12:04 |
Dyrcona |
It finished while I was up getting lunch ready. |
| 12:05 |
Dyrcona |
I think the unnest function is responsible for my NULL output. |
| 12:05 |
Dyrcona |
Laptop is warm.. Guess my fingers won't be cold much longer. |
| 12:13 |
Dyrcona |
oof. missing titles throws the column count off, but I have a record that I can look at to see if there are nulls in the rmsr.isbn array. I suspect yes. |
| 12:15 |
Dyrcona |
Well, no.... |
| 12:15 |
Dyrcona |
I bet some don't convert to isbn13 and that blows it up. |
| 12:15 |
Dyrcona |
I'll check on that later. |
| 13:27 |
Dyrcona |
Yeah, so invalid ISBNs come back null. Rather than try to improve the query, I'll filter the output through sed to delete lines beginning with \N. However, I've got a bigger issue that I might as well resolve now. Some of the copies don't show up in asset.opac_visible_copies, though they are visible in the OPAC, so looks like I need to overhaul the query anyway. |
| 13:36 |
Dyrcona |
Anyone know the magic sauce that is used by the OPAC to find visible copies to save me from having to pour over code? |
| 13:59 |
Dyrcona |
I feel like this shouldn't have been a surprise, like I should already know what to do about it. |
| 14:07 |
Dyrcona |
Really useful release note: "Public and staff catalog search is now both more accurate and faster by redesigning how the visibility of records is calculated." |
| 14:25 |
Dyrcona |
So, it's too complicated to do visibility testing in the database, now? |
| 14:29 |
Dyrcona |
Oh. I missed opac_visible on asset.copy. |
| 14:29 |
* Dyrcona |
tries that. |
| 14:31 |
Dyrcona |
I don't think that's as useful as it looks. |
| 14:37 |
Dyrcona |
Ah, OK. Much more to check. |
| 15:17 |
Dyrcona |
@monologue |
| 15:17 |
pinesol |
Dyrcona: Your current monologue is at least 29 lines long. |
| 15:25 |
* berick |
didn't want to interrupt :) |
| 15:25 |
Dyrcona |
heh. |
| 15:26 |
Dyrcona |
I think I've figured out my work around for the disappearance/lack of update of asset.opac_visible_copies. I'm basically copying the select from the function that maintained the table for now. |
| 15:27 |
Dyrcona |
I don't grok how to use the vis_attr_cache stuff just yet seeing as what I could find was Perl that builds the queries in bits and pieces. |
| 15:28 |
Dyrcona |
I now have a Pg question that I'll save for #postgresql on Tuesday. |
| 15:36 |
Dyrcona |
Query seems to be a bit slower now. I should run an explain analyze on it. |
| 15:45 |
Dyrcona |
It might be faster just to maintain asset.opac_visible_copies. :) |
| 15:53 |
Dyrcona |
It has been running for 50 minutes. I wonder if it will finish before New Year's... :) |
| 15:58 |
Dyrcona |
Using a good bit of temp space, too.... |
| 16:02 |
miker |
Dyrcona: I'm not here, but asset.copy_vis_attr_cache is the table you want in modern EG |
| 16:02 |
miker |
oh, you found that. nm |
| 16:15 |
Dyrcona |
Yeah, I haven't figured out how to use it, yet. |
| 16:16 |
Dyrcona |
My new version produces about 148 more entries than my previous version based on a refreshed asset.opac_visible_copies. I'll take that for now. |
| 16:17 |
Dyrcona |
Maybe I'll play asset.copy_vis_attr_cache for the next few minutes. |
| 16:20 |
* Dyrcona |
wonders if there's any documentation on vis_attr_vector. That was the bit I couldn't figure out from the code. It looks like entry 0 is opac_visible, but I don't know what the rest means. |
| 16:23 |
Dyrcona |
Meh. Too late in the day/week/month/year to worry about that. I'll look at it next week. |
| 16:24 |
miker |
there's some code comment documentation in the helper functions that build the int array, and that create the queryint value for testing against it, but, as you say, it's 30min from the end of the working year |
| 16:39 |
Dyrcona |
I'll see if I can find that next week. |
| 16:40 |
Dyrcona |
Happy New Year's everybody! |
