00:59 |
|
book` joined #evergreen |
04:23 |
|
mtj_ joined #evergreen |
04:41 |
pinesol_green |
Incoming from qatests: Test Failure - http://testing.evergreen-ils.org/~live/test.html <http://testing.evergreen-ils.org/~live/test.html> |
07:23 |
|
mrpeters joined #evergreen |
07:34 |
|
collum joined #evergreen |
07:44 |
|
rjackson_isl joined #evergreen |
08:29 |
|
dkyle joined #evergreen |
08:40 |
|
mmorgan joined #evergreen |
08:42 |
|
maryj joined #evergreen |
08:44 |
Stompro |
Happy test writing day everyone, Woohoo |
08:44 |
jboyer-isl |
miker: In that situation, wouldn’t the logs that would be backed up just be lost? That’s obviously preferable to knocking app servers offline, but still not ideal I would think. |
08:55 |
|
ericar joined #evergreen |
08:56 |
* csharp |
confirms that PINES is using UDP to ship logs |
09:34 |
miker |
jboyer-isl: not necessarily. depends on buffer settings at various levels |
09:36 |
jboyer-isl |
Ah, ok. |
09:51 |
rfrasur |
Soooo, does this mean no one is volunteering to set up the festival canopy? |
09:56 |
berick |
i think you can also configure rsyslog to drop new messages when its outbound queueu fills up instead of throttling the message submitter |
09:57 |
berick |
as a last layer of defense |
10:01 |
* berick |
puts on his test-writing pants |
10:04 |
ldw |
How do we want to work on tests today? Do people want to take a bug and work on it idependently? Do we need to discuss pgTap or test writing in general? |
10:04 |
ldw |
s/idependently/independently/ |
10:10 |
|
Christineb joined #evergreen |
10:13 |
|
Shae joined #evergreen |
10:14 |
Stompro |
I think I understand what needs to be done for the evergreen.lowercase test, I just have to get things setup so I can test it. I'll let you know if I run into anything. |
10:18 |
berick |
i'm going off-script a little bit and doing a perl live test. will return to pgtap after that. |
10:18 |
jboyer-isl |
I’ll be pretty in and out as far as test writing day goes ( :( ) but I would like to recommend that tests include things expected to fail. I don’t mean tests that fail, I mean writing the tests so that they basically say “This sql/function/etc. should fail and that’s ok. If it doesn’t fail, fail this test.” |
10:20 |
ldw |
jboyer-isl: do you mean pass in data to the function that should fail it and check for the fail? |
10:21 |
jeff |
that's how i interpreted his statement. |
10:22 |
jboyer-isl |
ldw: Yes. It depends on what’s be tested of course, it doesn’t make sense for every test. The thinking is that if the test starts failing, something has broken in the function. (it’s not validating arguments correctly, it’s breaking a contract that it will never return NULL, etc.) |
10:36 |
Stompro |
Debian Wheeze has a package for pgtap 0.90 released in 2011... I wonder if that will work? I'm getting errors " ERROR: function isnt_empty(unknown, unknown) does not exist" |
10:37 |
ldw |
Stompro: you need to run CREATE EXTENSION pgtap; on your database. |
10:37 |
ldw |
I am running into issues with pgTap and Postgres 9.4. I am going to fall back to 9.1 for today. |
10:40 |
Stompro |
I'm just running pg_prove, "pg_prove -U evergreen -d egdb -h localhost t" |
10:45 |
Stompro |
Do you install the pgtap extension for just your Evergreen db, or globally, for template1? |
10:46 |
ldw |
Stompro: your own database |
10:46 |
Stompro |
Running \dx shows "pgtap | 0.90.0 | evergreen | Unit testing for PostgreSQL |
10:46 |
Stompro |
" |
10:47 |
jboyer-isl |
Stompro: I believe things added to template1 only affect databases created after template1 is changed. Existing dbs are unaffected. |
10:47 |
Stompro |
jboyer-isl, makes sense, thanks. |
10:48 |
jboyer-isl |
The answer appears to be that pgtap .90 does not include isnt_empty, only is_empty: https://github.com/theory/pgtap/blob/v0.90.0/sql/pgtap.sql.in |
10:48 |
jboyer-isl |
That’s unfortunate. :( |
10:50 |
Stompro |
I'll switch to a debian jessie test system with a newer version, thanks for finding the answer jboyer-isl++ |
10:50 |
Dyrcona |
Stompro: You could try installing a more recent pgtap from source, just sayin'. |
10:52 |
Stompro |
I know, I just want to get going on writing the tests and not mucking with building pgtap today. |
11:01 |
Stompro |
Sigh, no debian package for pgtap in Jessie, I guess I will be building it. |
11:06 |
jeff |
looks like the pgtap package in debian is not well maintained. as such, it was removed from testing back in 2013. |
11:06 |
* berick |
grabs 1483508 |
11:06 |
dbs |
http://apt.postgresql.org/pub/repos/apt/pool/main/p/pgtap/ may be an option |
11:07 |
csharp |
I don't mean to distract from all the test-writing awesomeness, but can someone confirm this behavior on a 2.7+ system using built-in selfcheck?... |
11:07 |
* jeff |
waits for csharp's question to be about usernames/barcodes |
11:07 |
csharp |
jeff: yep ;-) |
11:07 |
dbs |
but then you're into the whole "use postgresql's apt repos instead of debian/ubuntu's" thing |
11:08 |
csharp |
with the Patron Barcode Format YAOUS set to ^[0-9]+$, when I enter a barcode or username, it appears to accept it, but doesn't accept the password after it's typed |
11:08 |
dbs |
http://wiki.postgresql.org/wiki/Apt fwiw |
11:08 |
csharp |
the web console shows that the regex works |
11:08 |
csharp |
testing <barcode#> ... barcode |
11:08 |
berick |
did we decide we can stop adding all of the \pset boilerplate at the top of these tests, since it's not needed with pg_prove ? |
11:09 |
csharp |
I see the ****** (dots) appear for the password, but Enter doesn't work |
11:10 |
csharp |
if I remove the regex from the YAOUS, it works fine |
11:10 |
Stompro |
Ok, building pgtap is about as simple as installing the package... I'm going to add a note to the qa wiki page. |
11:59 |
|
bmills joined #evergreen |
12:03 |
|
jihpringle joined #evergreen |
12:07 |
Stompro |
Is there a trick to pasting unicode characters into putty? I can view them fine, but I cannot copy them. |
12:07 |
pastebot |
"yboston" at 64.57.241.14 pasted "first pgTAP test attempt" (42 lines) at http://paste.evergreen-ils.org/24 |
12:08 |
yboston |
I was wndering if I could get some feedback on my first pgtap test. I am testing a function called public.first_agg from Open-ILS/src/sql/Pg/002.functions.aggregate.sql |
12:08 |
yboston |
LP 1483506 |
12:08 |
pinesol_green |
Launchpad bug 1483506 in Evergreen "public.first_agg needs test" (affected: 1, heat: 6) [Undecided,New] https://launchpad.net/bugs/1483506 - Assigned to Yamil (ysuarez) |
12:08 |
yboston |
what file name should I use? |
12:10 |
berick |
yboston: file name for the test? |
12:10 |
yboston |
berick: yes |
12:11 |
yboston |
also, not sure how the actual function is being used |
12:11 |
berick |
yboston: something like Open-ILS/src/sql/Pg/t/lp1483506-first-agg.pg should do |
12:11 |
yboston |
to help me test it better |
12:11 |
berick |
see other examples in Open-ILS/src/sql/Pg/t/ |
12:11 |
yboston |
OK, was not sure if I shuld put the LP number or not |
12:12 |
berick |
doesn't really matter, but may as well |
12:13 |
yboston |
the LP bug was created for todays "test day," the LP bug does not have any significant content |
12:13 |
berick |
so, first_agg() is a very simple function. pass in 2 things, it returns the first one that's not NULL. if they're both null, it returns NULL |
12:14 |
yboston |
sorry, I meant I was havign a hard time finding it used int he code. Might have grepped wrong |
12:14 |
yboston |
I am not sure if I shuld be testing with more than simple data types. not sure if I should test with passing arrays(?) |
12:14 |
berick |
hm, don't think it is used in the code. |
12:14 |
berick |
see no references to it |
12:15 |
yboston |
berick: thanks, I feel less crazy now |
12:15 |
yboston |
maybe in the past it was used, but with all the cleanign that has been done it is no longer needed |
12:16 |
berick |
maybe |
12:17 |
yboston |
berick: thanks for the feedback, I feel less crazy now. |
12:21 |
dbs |
berick: yboston: maybe no lp number in the filename, so that we can add more tests to the same file later? |
12:21 |
yboston |
dbs: makes sense |
12:22 |
yboston |
what about file name structure, shoudl I prepend the SQL file name then add the fucntion name? |
12:22 |
berick |
yeah, that makes sense. perhaps we should clean up the existing ones too |
12:22 |
berick |
re: no lp number |
12:23 |
berick |
yboston: hm, i would just do the function name |
12:23 |
yboston |
berick: good idea |
12:24 |
yboston |
can we use subfolders> |
12:24 |
yboston |
? |
12:24 |
yboston |
berick: I think first_agg gets used by calls to first() |
12:24 |
yboston |
if so, I have more ideas of how to test it |
12:25 |
berick |
yes, in fact it does |
12:25 |
yboston |
may add a test for first() too |
12:25 |
berick |
good catch |
12:27 |
Stompro |
Hmm, evergreen.lowercase doesn't seem to handle the two turkish I's correctly. |
12:28 |
yboston |
so the two ideas are soemthing like 1) 002.functions.aggregate.sql_first_agg.pg 2) first_agg.pg |
12:30 |
ldw |
I like subfolders. But pg_prove needs to be run with --recurse for that to work. |
12:30 |
yboston |
berick: makes sense |
12:30 |
yboston |
ldw: good to know |
12:30 |
berick |
i mean, if we're going to use the file name, we just need to also change the test name, if we're going to include the file name in the test |
12:30 |
berick |
should a function move |
12:31 |
berick |
maybe that's very rare, not worth worrying about |
12:31 |
berick |
anyway, it popped in my head, and i blurted sans filter :) |
12:31 |
yboston |
I think we don't have to come up with a solid decision right now, I just wanted to run it by soem folks to get the cnversation started |
12:32 |
berick |
hm, i should add an aggregate test to my text_concat test as well |
12:33 |
csharp |
okay - I can confirm the same "set Patron Barcode Format setting and bork selfcheck" issue on a test server running master |
12:33 |
* csharp |
runs off to open the bug report |
12:34 |
yboston |
BTW, since pg_proves needs --recurse for subfolders to work, I will avoid using a subfolder |
12:34 |
yboston |
I look forward tpo having so many tests that we need to start using subfolders |
12:35 |
berick |
yboston++ # fixing problems before they become problems |
12:37 |
yboston |
berick for your feedback |
12:37 |
yboston |
berick++ |
13:53 |
dbwells |
Right, it only affects circs where there is an ungenerated fine at checkin, and the balance was zero before checkin. |
13:53 |
bshum |
Oh nevermind, I see you assigned yourself. |
13:53 |
bshum |
You're on top of it |
14:39 |
phasefx |
dbwells: unrelated, but in the same area, it looks like that having that lost/zero setting could prevent fines from being generated at all during checkin |
14:40 |
* phasefx |
hasn't actually tested that yet though |
14:41 |
dbwells |
phasefx: Thanks. I've got a fix for the first part (just writing a test), but I'll need to think more on that. The idea there is to purposely not generate fines if it is a lost checkin with currently zero fines, as that is what I understand the setting to mean. |
14:42 |
bshum |
dbwells: yeah I think that's the new setting that miker added for keeping stuff at zero. Stop gap for negative balance stuff :P |
14:43 |
phasefx |
dbwells: I did put some logging statements in testing the reported bug, and the handle_fines that gets called is the one inside if (!$dont_change_lost_zero) { |
14:44 |
* phasefx |
squints, okay, it's probably safe, since !$dont_change_lost_zero isn't just based on the setting, but looks like it considers the actual copy status, etc |
14:45 |
|
buzzy joined #evergreen |
14:45 |
dbwells |
right. It tries to just short-circuit the whole thing, and I think it's right, or at least close ;) |
14:53 |
bshum |
Okay... |
15:00 |
bshum |
Now of course, I have more users than the list i was given, but I can work with that. |
15:00 |
bshum |
dbwells++ # thanks! |
15:01 |
dbwells |
money.billable_xact is the "parent" table of action.circulation, money.grocery, and booking.reservation, so they all share a pool of ids. |
15:04 |
bshum |
After we remove or void those entries, part of me thinks I should do something to refresh the user summaries |
15:05 |
bshum |
Ah, there's a trigger on money.billing for that sort of thing, maybe... |
15:06 |
* bshum |
checks |
15:06 |
bshum |
(and tests) |
15:14 |
bshum |
Oh happy day. |
15:21 |
jboyer-isl |
Bmagic: if you are still looking into ways to relieve rsyslog congestion, look up the $ActionQueueType, $ActionQueueFileName, $SystemLogRateLimitInterval, and $SystemLogRateLimitBurst directives. |
15:22 |
jboyer-isl |
On a related note, fulling turning on that firehose for our database is yielding about 20%+ larger logs than letting the rate limiter eat thousands of what are likely 0ms queries. |
15:25 |
ldw |
jboyer-isl: I am writing negative tests, but I am using the isnt function to make sure they pass. This results in all tests passing. Does this seem reasonable to you, or is there a specific reason you had for wanting a test to fail? |
15:26 |
jboyer-isl |
ldw: that’s exactly what I meant. if you just run pg_prove it should say all tests pass. A test that passes because it’s expecting a function to fail still passes. :) |
15:26 |
ldw |
ok cool :) |
15:39 |
Stompro |
ldw, I'm looking at the pgtap test for change_db_setting , which calls alter database... but it looks like changes to alter database only take effect for new sessions. Which would be outside the testing transaction, so I don't think that is testable. What do you think? |
15:39 |
Bmagic |
jboyer-isl: thanks! for now, I am going to leave at local logging. I am afraid of bringing it all down again. Makes for an unpleasant day! |
15:39 |
ldw |
Stompro: you should enclose your tests in a BEGIN and ROLLBACK block, so the changes take effect for the duration of the test, but are not permanent. |
15:40 |
|
jlitrell joined #evergreen |
15:40 |
ldw |
Stompro: I see what you are saying |
15:40 |
ldw |
hmmm |
15:40 |
jeff |
evergreen.change_db_setting is involved in the search_path mess. |
15:40 |
Stompro |
jeff, that is the only thing it is used for that I can see. |
15:41 |
jeff |
i believe it is only used to change the db search_path, and potentially in a way that doesn't account for the fact that such changes are not immediate. i haven't dug into it further/recently. |
15:41 |
jeff |
but i remember it from my search_path digging. |
15:41 |
Stompro |
So I was trying to change the search path, and then show the search path, but if you want to change the search_path for the current session you use 'set' , not alter db. |
15:41 |
* jeff |
nods |
15:42 |
jeff |
if pg_tap supports disconnecting and reconnecting between test steps, that might be useful. if not, then this might not be a great function to test Right Now. |
15:43 |
Stompro |
But would the transaction persist across connections? |
15:43 |
jeff |
no. |
15:44 |
ldw |
could it be SELECTed from the SCHEMA instead of specifying a value to expect, specify a SQL statement? I am not sure if you can do that. |
15:44 |
Stompro |
and the change would rollback at the disconnect. So lets skip that one. |
15:45 |
Stompro |
Maybe, I'll dig and see where those values are stored. |
15:47 |
ldw |
Stompro: I am not sure if the second paramter from the tests can be a SQL statement. I might be wasting your time if you are digging through the schema. I am trying to see if I can get a SELECT statement to run as the second paramter. But, no luck so far. |
15:47 |
Stompro |
I was using results_eq Which takes a query and what the results should match. |
15:48 |
ldw |
you could run a \set after the alter to set a variable to the value of a select statement and use that value in your test. |
15:59 |
ldw |
Stompro: hmm, I cannot get \set to equal a value of a SELECT statement either. |
02:16 |
|
jonadab_znc joined #evergreen |
03:22 |
|
gsams joined #evergreen |
03:24 |
bshum |
@later tell dbwells Curious to know if you had any time to poke at https://bugs.launchpad.net/evergreen/+bug/1379815. We've been thinking about adding copy stat cats more permanently into some of our workflows, but can't if Vandelay doesn't support something like it. |
03:24 |
pinesol_green |
bshum: The operation succeeded. |
03:24 |
pinesol_green |
Launchpad bug 1379815 in Evergreen "Assign stat cats during Vandelay import/overlay of items" (affected: 1, heat: 8) [Wishlist,New] - Assigned to Remington Steed (rjs7) |
03:24 |
|
jcamins joined #evergreen |
03:43 |
|
gsams joined #evergreen |
03:43 |
|
Mark__T joined #evergreen |
03:47 |
|
dbs joined #evergreen |
03:53 |
|
mnsri_away joined #evergreen |
04:50 |
|
Mark__T joined #evergreen |
05:16 |
pinesol_green |
Incoming from qatests: Test Success - http://testing.evergreen-ils.org/~live/test.html <http://testing.evergreen-ils.org/~live/test.html> |
06:14 |
|
rlefaive joined #evergreen |
07:15 |
|
artunit_away joined #evergreen |
08:02 |
|
gsams joined #evergreen |
11:56 |
Bmagic |
once I removed that config and restarted rsyslog on the app servers, the log file stopped adding those error messages |
12:00 |
jeff |
logical. your app servers were trying to log to files that they didn't need to log to, and their errors about not being able to do so were being logged on the target/remote log host. :-) |
12:01 |
Bmagic |
jeff: I think that was it, time will tell of course |
12:02 |
Bmagic |
jeff: We have been having weird issues and these logs might be a clue. Every week for the last 4 weeks, ALL of our app servers get taken out of LVS rotation because apache stops responding to the test from lvs |
12:02 |
Bmagic |
Do you think that an rsyslog misconfiguration could possible pile up high enough to bring apache down to a slow enough crawl? |
12:03 |
bshum |
mrpeters: What do you mean when you say xpath issues? |
12:04 |
mrpeters |
ive done various ways of doing pg dump and it seems like i had some kind of issue with xpath or search path or something like that |
13:31 |
Christineb |
I can't stress how happy this made me :D |
13:31 |
jihpringle |
it's slated for 2.9 but we've backported to 2.8 and have been using it since July |
13:31 |
mcrawley |
don't know what that one is? |
13:33 |
lstjohn |
we are also about to sign off on #1402018 workstation scoping - we've tested it on 2.7.2 |
13:33 |
jihpringle |
the acquisitions admin menu was moved to be under the Admin menu rather than under Admin -> Server Administration |
13:33 |
Christineb |
The Acquisitions Admin menu has been moved directly under Admin instead of having to to into Admin --> Server Admin |
13:33 |
Christineb |
lstjohn - that is great news!! |
13:40 |
lstjohn |
I have a question. When you delete a line item from a purchase order that has not been activated, it deletes the line item from the selection list but not from the OPAC - is this your experience also? |
13:40 |
Christineb |
lstjohn - I have seen this happen |
13:41 |
Christineb |
when a MARC order file is loaded in "Load MARC Order Records" and Load Items for Imported Records is selected |
13:42 |
Christineb |
Is on my to-do list to do some more testing to figure out if this is a bug or desired behaviou |
13:42 |
Christineb |
does this scenario match what you are seeing? |
13:44 |
lstjohn |
Yes, order records loaded, copies created - there seems to be a related bug https://bugs.launchpad.net/evergreen/+bug/1269574 |
13:44 |
pinesol_green |
Launchpad bug 1269574 in Evergreen "ACQ lineitems canceled via EDI not deleting linked bibs/items" (affected: 5, heat: 26) [Medium,Confirmed] |
15:29 |
Dyrcona |
jeff csharp bshum: lp 1484655 |
15:29 |
pinesol_green |
Launchpad bug 1484655 in Evergreen "ftp://ftp.mozilla.org has moved to http://archive.mozilla.org" (affected: 1, heat: 6) [High,New] https://launchpad.net/bugs/1484655 - Assigned to Jason Stephenson (jstephenson) |
15:29 |
Dyrcona |
Oops. Should unassign myself. ;) |
15:30 |
jeff |
say, why didn't live tests break? because it doesn't build a staff client? |
15:31 |
Dyrcona |
I successfully built a client last week, so don't know when this actually changed. |
15:31 |
Dyrcona |
Also, maybe it doesn't build a client. |
15:32 |
jeff |
timestamp on the README in question is "8/7/15, 10:22:00 PM" |
15:33 |
Dyrcona |
Missed it by a couple of days. |
16:25 |
|
ericar joined #evergreen |
16:35 |
|
jihpringle joined #evergreen |
16:59 |
pinesol_green |
Incoming from qatests: Test Success - http://testing.evergreen-ils.org/~live/test.html <http://testing.evergreen-ils.org/~live/test.html> |
17:01 |
Dyrcona |
jeff: Looking through the test logs above, it doesn't appear to make a client. It just does the server-side bits. |
17:01 |
* Dyrcona |
disappears. |
17:08 |
|
mmorgan left #evergreen |
17:32 |
|
jlundgren left #evergreen |
18:00 |
dbwells |
rfrasur: what's up? |
18:02 |
rfrasur |
I think Rogan is taking care of it. Maybe? Thank you :-) |
18:02 |
dbwells |
okay :) |
18:04 |
berick |
Bmagic: yes, i think requesting a static file from your apache server via http is a common way to confirm it's alive |
18:04 |
berick |
Bmagic: many load balancers use that as a way to take server in/out of rotation. request a file (e.g. /openils/var/web/ping.tx) and if it's not there, the server is removed from rotation |
18:05 |
berick |
well, ping.txt (but i imagine it can be anything) |
18:06 |
berick |
that style of test is great because you can take a server out w/o having to touch the load balancer |
18:37 |
rfrasur |
Okay, I'm thinking through some stuff for our library. In a consortium...or library system...can you hide all the collections of a branch from the rest of the library system? |
18:38 |
rfrasur |
I mean, using Evergreen...not a book truck and a closet. |
18:41 |
|
buzzy joined #evergreen |
02:23 |
|
Mark__T joined #evergreen |
04:48 |
|
Mark__T joined #evergreen |
05:12 |
|
artunit_ joined #evergreen |
06:43 |
csharp |
ldw++ # test wrangling |
07:37 |
|
Callender joined #evergreen |
07:45 |
|
TaraC joined #evergreen |
07:46 |
|
mrpeters joined #evergreen |
08:50 |
krvmga |
in eg 2.5 it was easy to change the order of search format filters in the opac dropdown list. since 2.6, they have been listed alphabetically. is it possible to change the order of the filters somehow other than by alphabetical manipulation? |
08:56 |
|
mmorgan joined #evergreen |
09:14 |
|
jwoodard joined #evergreen |
09:25 |
Stompro |
mmorgan++ thanks for testing the patron expiration notice.. now to get the release notes done. |
09:25 |
|
yboston joined #evergreen |
09:27 |
* Dyrcona |
wonders if hopkinsju had any success with backstage yesterday evening. |
09:27 |
mmorgan |
Stompro++ Thanks for polishing the code, and (in advance) for the release notes! |
09:29 |
|
ericar_ joined #evergreen |
09:35 |
|
collum joined #evergreen |
10:07 |
|
pmurray_away joined #evergreen |
10:10 |
Stompro |
ldw++, thanks for all the test writing bugs, I assigned myself to the simplest one I could find :-) |
10:23 |
krvmga |
is there any way to change the order of search format filters in the opac other than by playing with the alphabetical listing? |
10:29 |
dbwells |
krvmga: I don't think there is anything built in. We ordered ours as you are thinking by manipulating the strings. In our case, we prepended a certain number of spaces to the "Search Label"s. The spaces do not show up. |
10:29 |
krvmga |
dbwells: can i look at your catalog url? |
14:30 |
Bmagic |
I was having trouble finding a report about the in-house barcode box not clearing the box after the enter button was pressed (ready to accept the next barcode) |
14:39 |
Bmagic |
https://bugs.launchpad.net/evergreen/+bug/1483857 |
14:39 |
pinesol_green |
Launchpad bug 1483857 in Evergreen "Web based staff client in house use barcode box clear" (affected: 1, heat: 6) [Undecided,New] |
14:59 |
ldw |
Stompro: thank you for taking a bug. Have you worked with the pgTap testing suite before? |
15:00 |
Stompro |
ldw, Never, All I know about it is from reading the docs that you linked to previously. |
15:00 |
|
rlefaive joined #evergreen |
15:03 |
ldw |
Stompro: If you have any questions about it do not hesitate to ask. |
15:04 |
|
akilsdonk joined #evergreen |
15:17 |
jboyer-isl |
I don't think I've asked this before, but what are the thoughts re: tests and simply adding new seed data? Don't bother, check to see if a few values are present, check every value, etc? |
15:18 |
berick |
jboyer-isl: for example? |
15:18 |
jboyer-isl |
I'm looking at LP #1371647 and have a large upgrade script ready to go, but I haven't decided what to do about a test case. |
15:18 |
pinesol_green |
Launchpad bug 1371647 in Evergreen "config.marc21_ff_pos_map needs an audit" (affected: 1, heat: 6) [Undecided,New] https://launchpad.net/bugs/1371647 |
15:20 |
jboyer-isl |
berick: example incoming |
15:21 |
jboyer-isl |
Bah, it would have, but I mistyped my nick... |
15:22 |
berick |
sorry, i was confuzzled |
15:23 |
berick |
i've just been checking a few values, like a basic sanity check. |
15:23 |
berick |
don't think we have a policy on that |
15:25 |
jboyer-isl |
Excellent. I don't mind spot checking a few things to be sure the patch was applied, but a "full check" would be a huge pain to write for things like that. |
15:26 |
jboyer-isl |
I'll try to get something put together and posted for proper discussion/testing asap; this is holding up some format search improvements that we're trying to do and I don't want it to be a local only fix. |
15:27 |
Dyrcona |
Well, you could use a macro in your text editor to write the tests for you... I've done that recently from other upgrade scripts. |
15:30 |
berick |
you're gonna need a unit test on that macro |
15:32 |
Dyrcona |
You're gonna need a unit test on your unit test, too. :) |
15:33 |
Dyrcona |
But actually, I didn't use a macro just a regex search and replace. I misspoke. |
15:45 |
jboyer-isl |
That's true, but then there's only full coverage for a chunk of data. If there's ever interest in fully verifying seed data, I'd rather see one big test that checks it all (one file per schema/table, not necessarily the entire seed database) |
15:52 |
Dyrcona |
jboyer-isl: That sounds reasonable to me. When do you start working on it? :P |
15:52 |
Dyrcona |
j/k |
15:54 |
jboyer-isl |
Well, I suppose someone could find and replace 950.config.seed-data.sql, heh. |
16:12 |
jlitrell |
Living in soup. |
16:21 |
Dyrcona |
Hmm. Looking into something from yesterday that I thought would require a staff client change and now it looks like I can handle it all on the back end. |
16:22 |
Dyrcona |
berick: Do you remember the LP bug that was created for the branch that allowed pre cat copies to be created for ILL and other purposes? |
16:22 |
Dyrcona |
kmlussier and I tested it, but I can't find the bug now, even search fix committed bugs. |
16:23 |
|
mmorgan1 joined #evergreen |
16:23 |
|
mmorgan1 left #evergreen |
16:24 |
|
mmorgan2 joined #evergreen |
16:24 |
pinesol_green |
Launchpad bug 1308239 in Evergreen "Support targeting and fulfillment of precat copy holds (for ILL)" (affected: 2, heat: 10) [Wishlist,Fix released] https://launchpad.net/bugs/1308239 |
16:24 |
|
mmorgan2 left #evergreen |
16:25 |
Dyrcona |
berick++ # that's a handy feature. |
16:26 |
Dyrcona |
Looks like bshum tested it, too. |
16:27 |
hopkinsju |
Dyrcona: Just made a breakthrough. |
16:27 |
Dyrcona |
hopkinsju: What was the problem? |
16:28 |
hopkinsju |
Initial issue was the default last_run_date and the fact that we didn't use the script to do the initial export |
17:00 |
Dyrcona |
Well, time to go home. |
17:12 |
|
mmorgan1 joined #evergreen |
17:12 |
|
mmorgan1 left #evergreen |
17:18 |
pinesol_green |
Incoming from qatests: Test Success - http://testing.evergreen-ils.org/~live/test.html <http://testing.evergreen-ils.org/~live/test.html> |
18:20 |
|
rjackson_isl_ joined #evergreen |
18:20 |
|
mmorgan joined #evergreen |
18:23 |
|
TaraC joined #evergreen |
10:07 |
Dyrcona |
csharp: We don't seem to have set one. |
10:08 |
Dyrcona |
We often get requests to fix patrons after registration, but it's usually when the staff tries to use a barcode that got assigned as the usrname on a previous user. |
10:08 |
|
collum joined #evergreen |
10:08 |
csharp |
Dyrcona: we discovered the need for it when testing the built-in selfcheck, which says "check whether the string the patron has entered matches the opac.barcode_regex setting... if not, treat the string like a username" |
10:09 |
csharp |
since our setting is null, all entered values are being treated like usernames |
10:09 |
Dyrcona |
Well, our opac.barcode_regex setting is a bit strange. It doesn't check length, for one thing. |
10:09 |
Dyrcona |
I thought you were asking about a different setting. |
10:10 |
Dyrcona |
But, the patron's barcode and usrname not matching, but both being barcodes, seem to happen in registration when the staff start out using 1 barcode and end up issuing a different one. |
10:41 |
|
Christineb joined #evergreen |
10:56 |
jeff |
interruptions, possibly. |
11:11 |
* mmorgan |
reads backscroll |
11:12 |
pinesol_green |
Incoming from qatests: Test Failure - http://testing.evergreen-ils.org/~live/test.html <http://testing.evergreen-ils.org/~live/test.html> |
11:12 |
mmorgan |
Our opac.barcode_regex specified the barcode must have at least five digits, that's all. Not sure where it actually comes into play. |
11:14 |
Dyrcona |
The test failure looks like one of those things we can't control. |
11:14 |
Dyrcona |
Got a 503 while trying to download the phantomjs package. |
11:17 |
phasefx |
start local repos for everything? |
11:27 |
Dyrcona |
Nope. It'll probably succeed when the issue on the server is discovered and resolved. |
15:26 |
|
jlitrell joined #evergreen |
15:27 |
jeff |
ah. title hold on a bib whose only volume is deleted. |
15:33 |
jeff |
and we have a surprising number of those. |
15:35 |
pinesol_green |
[evergreen|Michael Peters] LP#1394989: Make users_of_interest test for defined actor.usr.card values - <http://git.evergreen-ils.org/?p=Evergreen.git;a=commit;h=13da6b8> |
15:35 |
pinesol_green |
[evergreen|Bill Ott] LP#1394989: Do not include deleted users when retrieving for Collections - <http://git.evergreen-ils.org/?p=Evergreen.git;a=commit;h=0f8ec14> |
15:39 |
|
berowskim joined #evergreen |
15:57 |
Bmagic |
I have a bib that is sorted to the bottom of the list search when sorted newest to oldest. Come to find out, there is no entry in metabib.record_sorter for pubdate for this bib. When I reingest the bib, it still doesnt get created. Where is the glue? |
16:02 |
Bmagic |
I found it. sorry, that was easy, Date1 in the 008 |
16:11 |
Dyrcona |
It handles authority deletes from a file matching DEL |
16:12 |
Dyrcona |
Anything else that doesn't begin with R followed by 2 digits is treated as an authorities file. |
16:13 |
Dyrcona |
We get zips with bibs, authorities, and reports all in one and it handles them OK. |
16:13 |
hopkinsju |
So if I have a zip file with ONLY a file called BIB.MRC in it, shouldn't that get overlayed? |
16:14 |
hopkinsju |
I'm trying to do some smaller scale testing before i set it loose. |
16:15 |
Dyrcona |
It depends. If you look in Backstage::Import->do_bibs you'll see there are some criteria, but you should get some output to the screen as each record is processed. |
16:15 |
Dyrcona |
Sorry, it's doBibs not do_bibs. |
16:15 |
hopkinsju |
Ok |
17:00 |
Dyrcona |
It's probably because of the full path to BIB.MRC in the zip file. |
17:01 |
Dyrcona |
The zip files that I get from Backstage have no directory information in the member file names. |
17:02 |
Dyrcona |
Anyway, signing off for now. |
17:03 |
pinesol_green |
Incoming from qatests: Test Success - http://testing.evergreen-ils.org/~live/test.html <http://testing.evergreen-ils.org/~live/test.html> |
17:05 |
|
mmorgan left #evergreen |
17:14 |
jeff |
deleted bibs with non-deleted call numbers. deleted call numbers with non-deleted copies. |
17:14 |
jeff |
always fun to try to determine which way to "fix" this. |
00:58 |
phasefx |
@later tell Dyrcona the live tests happen on a near pristine snapshot of wheezy and runs all the tests |
00:58 |
pinesol_green |
phasefx: The operation succeeded. |
02:53 |
|
sarabee joined #evergreen |
03:27 |
|
gsams joined #evergreen |
09:05 |
kmlussier |
And I see rain in the forecast for Tuesday and Wednesday. Figures. |
09:13 |
Stompro |
mmorgan, have you started on the release notes for bug 1124498, I also started on that but forgot to assign it. |
09:14 |
pinesol_green |
Launchpad bug 1124498 in Evergreen "Wishlist: Patron notification via email when card is about to expire" (affected: 6, heat: 40) [Wishlist,Confirmed] https://launchpad.net/bugs/1124498 - Assigned to Michele Morgan (mmorgan) |
09:15 |
mmorgan |
Stompro: I was actually working on final testing to sign off. You are welcome to the release notes :) |
09:18 |
Stompro |
mmorgan, Wonderful, I hope your testing goes well. |
09:18 |
kmlussier |
mmorgan++ Stompro++ |
09:19 |
kmlussier |
Stompro: I think MVLC might do some of the things you described in your email with staff clients for test servers. tsbere might be able to answer your questions. I don't think he's around this week. |
09:21 |
* mmorgan |
is also interested in that info. I often find a need to run 2 clients simultaneously on the same workstation. |
09:21 |
kmlussier |
I use MVLC's servers quite a bit for testing. Icons for the clients that access the test servers are a different color and they don't share settings with the other clients on my local machine. |
09:21 |
* kmlussier |
has many, many clients on her local machine. |
09:22 |
Stompro |
kmlussier, great, any guidance would be appreciated. I'll send him an email if I don't hear from him. |
09:22 |
kmlussier |
mmorgan: Yes, that's what I like about testing on MVLC's servers. I can have two or more clients at the same time. tsbere also configured the clients for the MassLNC VMs to work the same way. |
09:23 |
kmlussier |
tsbere++ #for general wizardry |
09:24 |
mmorgan |
Yes, I noticed that. I can run a MassLNC client concurrently with a NOBLE client, but not a NOBLE Production along with a NOBLE training client. |
09:25 |
mmorgan |
Stompro: We will just change the hostname when the different servers are running the same version of Evergreen, but often our servers are on different versions. |
09:27 |
Stompro |
mmorgan, I think you can run multiple instances at the same time by enabling multi profile mode and multi-instance mode - http://wiki.evergreen-ils.org/doku.php?id=mozilla-devel:building_the_staff_client#multiple_workstations_on_one_install |
09:28 |
Stompro |
Add -P and -no-remote to your client shortcut. |
09:30 |
mmorgan |
Stompro: we use -profilemanager all the time. I'll have a look at -no-remote, though. |
09:34 |
Stompro |
I just wanted to make it super easy for staff to connect to the training/testing server, without needing to change the hostname, to take out a step where mistakes can be made. I'm just worried about staff accidently thinking they are are production when they are on testing and vice versa. Maybe adding a prefix to all usernames on the test system would help. |
09:35 |
mmorgan |
Stompro++ You made my day! |
09:42 |
Stompro |
Did the -no-remote work? |
09:44 |
mmorgan |
Yes! I now have 2 clients open at the same time :) Not sure how I missed that detail, but am happy to know it now! |
10:41 |
Bmagic |
Ok, that is whay I thought, just changing the parent OU should do it with autogen. Sounds easy enough |
10:42 |
Bmagic |
/whay/what |
10:42 |
Dyrcona |
You might have some memcached issues, so I'd wait and do it in the middle of the night when you can restart memcached without wrecking too much. |
10:45 |
Dyrcona |
I'd also suggest trying it on a test/training system first, just to make sure. |
10:50 |
* kmlussier |
is spending more time re-learning AsciiDoc conversion than writing documentation. |
10:50 |
csharp |
Bmagic: yes, we've moved a branch from one parent to another - it's not a big deal - just move it and run autogen (though you might schedule it after business hours) |
10:51 |
csharp |
all holdings and patrons, etc. remain intact |
11:03 |
Dyrcona |
My guess is that might be a mistake. |
11:04 |
kmlussier |
Yes, I don't think that was the intent at all. |
11:04 |
Dyrcona |
But, I'd need to play with with some samples and don't have time at the moment. |
11:04 |
Bmagic |
kmlussier: yep! That's what it looks like. I created a new 700 field on a test record and only created a subfield "a" and violla it was indexted |
11:05 |
Bmagic |
I am assuming that the contents of config.xml_transform are lifted from some global standard? |
11:11 |
Dyrcona |
Bmagic: http://www.loc.gov/standards/mods/ |
11:11 |
Bmagic |
Dyrcona: right, and the xsl structure was downloaded somewhere or did the EG community create it from their guideline? |
13:21 |
kmlussier |
Dyrcona: No. |
13:22 |
kmlussier |
Bmagic: I added a comment to https://bugs.launchpad.net/evergreen/+bug/1331174/comments/15 |
13:22 |
pinesol_green |
Launchpad bug 1331174 in Evergreen "Long Overdue processing needs org unit settings separate from Lost Processing" (affected: 4, heat: 18) [Wishlist,Confirmed] |
13:23 |
kmlussier |
Bmagic: I don't know what kind of time you have today, but if you were able to get that branch ready today, I could load it on a test server before I leave for vacation. |
13:23 |
kmlussier |
mmorgan is interested in testing it. |
13:25 |
* kmlussier |
is trying to get various billing bits in before she tackles billing documentation for 2.9. |
13:25 |
kmlussier |
Which brings me to bug 1436797. Maybe that's one I can handle. |
13:25 |
pinesol_green |
Launchpad bug 1436797 in Evergreen ""grocery" term confuses frontline staff and patrons" (affected: 11, heat: 48) [Undecided,New] https://launchpad.net/bugs/1436797 |
13:41 |
Stompro |
dyrcona++ Thanks for enlightening me about the STAFF_CLIENT_NAME variable, that is just what I needed. I'll add that to the buidling the staff client wiki page. |
13:42 |
Stompro |
Dyrcona++ (getting the caps right) |
14:12 |
pinesol_green |
csharp: It is possible. |
14:12 |
Dyrcona |
Stompro: I think it might use the stamp id as the product tag in windowssetup.nsi, but I never actually install the Windows staff client, so I don't know for sure. |
14:14 |
Stompro |
I don't thinks so, my STAMP_ID is set to rel_2_8_2 and the PRODUCT_TAG is set to '2.8' which gets combined with "Evergreen Staff Client" to set the default install. |
14:17 |
Stompro |
Dyrcona, I just tested it out, and setting a custom PRODUCT_TAG in windowssetup.nsi does set the default install directory. So I think I've got all the pieces figured out. Thanks for your help. |
14:18 |
Stompro |
I love having all the various options on how this stuff can work. |
14:18 |
Stompro |
iii-- |
14:21 |
Dyrcona |
Stompro: Cool! Glad you find my ramblings useful. |
16:47 |
kmlussier |
Heading out for the week. Have a nice week everyone! |
16:48 |
mmorgan |
Have a great vacation! |
16:49 |
|
kmlussier left #evergreen |
17:07 |
pinesol_green |
Incoming from qatests: Test Success - http://testing.evergreen-ils.org/~live/test.html <http://testing.evergreen-ils.org/~live/test.html> |
17:08 |
|
mmorgan left #evergreen |
17:17 |
|
mrpeters left #evergreen |
18:09 |
|
rlefaive joined #evergreen |
02:12 |
|
jonadab_znc joined #evergreen |
05:09 |
pinesol_green |
Incoming from qatests: Test Success - http://testing.evergreen-ils.org/~live/test.html <http://testing.evergreen-ils.org/~live/test.html> |
06:40 |
|
rlefaive joined #evergreen |
07:03 |
|
rlefaive joined #evergreen |
07:25 |
|
rlefaive joined #evergreen |
09:41 |
Dyrcona |
Dunno, bshum. You tell me. :) |
09:42 |
bshum |
Dyrcona: Oh I was just musing on dbs' musings |
09:42 |
Dyrcona |
Yep. |
09:43 |
* Dyrcona |
goes back to writing a pgtap test for jeffdavis_afk |
09:43 |
Dyrcona |
Well, one of his branches, anyway. |
09:43 |
bshum |
:P |
09:44 |
Dyrcona |
But, bshum's latest comment on the bug he mentioned might prove helpful to dbs if that is the problem. |
10:10 |
bshum |
But not to master |
10:10 |
bshum |
And since translations in LP track against master |
10:10 |
bshum |
Those strings never showed up for needing translation in LP. |
10:10 |
Dyrcona |
miker: If you want sprint2 in master before the alpha/beta release, then you'll need to clean up the upgrade scripts and provide pgtap tests. |
10:11 |
Dyrcona |
miker: I'm also concerned about the FIXME comment in XXXX.schema.marc-tag-tables.sql |
10:11 |
Dyrcona |
I get errors on line 9 and will try reordering the drop statements, but also wonder if I should even be running those drops given the comment. |
10:12 |
phasefx |
bshum: fun |
10:13 |
bshum |
phasefx: I would imagine we could still cherry-pick berick's changeset from 2.8 to master now. And then when Dyrcona does i18n dances for 2.9, we just make sure to bring in all the translations |
10:13 |
bshum |
But I'm not sure how we can fix 2.8. Yet. |
10:13 |
bshum |
Still rethinking it over. |
10:13 |
Dyrcona |
I should do a test release build whether there is an alpha or not. |
10:14 |
bshum |
Dyrcona: It's always good practice. |
10:15 |
Dyrcona |
Yep. Trouble is: not much time for practice. |
10:16 |
phasefx |
bshum++ I'm glad you're thinking about it; it's over my head at the moment. So we do something to extract strings from source files into language files, commit those somewhere, and Launchpad can then change them or let us export a new updated set of files based on its global translation effort? |
10:27 |
bshum |
That should go on a t-shirt |
10:28 |
* csharp |
copyrights it before bshum gets a chance |
10:30 |
Dyrcona |
Hmm... Wonder if IRC chat is copyrightable.... That would be an interesting case. |
10:31 |
Dyrcona |
And, I still haven't written the pgtap test, yet. |
10:32 |
dbs |
Your chat in IRC is an expression of an idea, thus very likely copyrightable. |
10:32 |
* dbs |
was just answering a question around copyright and fair dealing for our online learning folks |
10:32 |
RoganH |
Yep. If you create the speech you can copyright it. You're just exercising your own right to perform it in public. |
10:38 |
Dyrcona |
(C) is not valid, so in Brazil, what you just said is in the public domain, csharp. :) |
10:39 |
bshum |
I'm comparing our outputs now, for giggles |
10:39 |
Dyrcona |
In Brazil it must have the c in the circle symbol, the word copyright is not good enough by itself. (That I remember from research from a number of years ago.) |
10:40 |
Dyrcona |
But, anyway.... that pgtap test... |
10:40 |
* bshum |
makes an adjustment to the query |
10:40 |
bshum |
Since he realizes current_shelf_lib = 22 doesn't mean the same thing in his system |
10:41 |
Dyrcona |
bshum: Try my database and use current_shelf_lib = 17, pretty much guaranteed to blow up at our busiest library. :) |
10:42 |
bshum |
I still got a return about 2533 ms |
10:42 |
bshum |
So 10 times faster |
10:42 |
bshum |
Chose our largest library |
10:42 |
csharp |
bshum: my current experiment is deleting (aka, "aging") most of the hold_request table on a test server - gonna see if that helps. |
10:43 |
bshum |
One of our busiest ILL libs, got 2977 ms |
10:43 |
bshum |
So yeah |
10:43 |
bshum |
The shape of the query is different |
10:44 |
Dyrcona |
Speaking of slow queries.... |
10:45 |
Dyrcona |
bshum and berick do you think that lp 1479953 should be backported to 2.8 and 2.7 as a bug fix? |
10:45 |
pinesol_green |
Launchpad bug 1479953 in Evergreen "Deleting queues containing many records is slow, can time out" (affected: 2, heat: 10) [Undecided,New] https://launchpad.net/bugs/1479953 |
10:46 |
Dyrcona |
That's what I'm trying to write the pgtap test for. |
10:46 |
bshum |
Dyrcona: Remind me what happens if you try to create an index where one already exists? |
10:46 |
Dyrcona |
The create index statement dies. |
10:47 |
Dyrcona |
But it might succeed if the names are different. |
11:09 |
berick |
bshum: arg, feel bad I caused a problem I still don't understand |
11:15 |
bshum |
berick: It's okay, I'm not sure I will ever fully understand it. |
11:21 |
bshum |
berick: When I can think through everything a bit more, I'll try making some changes to that release_process wiki page on how the i18n bits happen |
11:22 |
Dyrcona |
Hmm. Should a commit adding a pgtap test and simply renaming indexes from the previous commit get another sign off? |
11:22 |
|
kbutler joined #evergreen |
11:22 |
berick |
bshum: k. let me know if I can help |
11:24 |
bshum |
berick: Looking back in the git history, I can see that changes for "po files" and "newpot" occurred as separate commits. |
11:39 |
bshum |
So yeah, every time we update the pot's in master, it'll setup new strings needing translation in LP |
11:40 |
bshum |
And then as people translate them in LP, it gets pushed to dbs' branch |
11:40 |
bshum |
Which we then pull back in when we run the updates_pofiles script |
11:46 |
pinesol_green |
[evergreen|Jeff Davis] LP#1479953: Add indexes to vqbr foreign key references - <http://git.evergreen-ils.org/?p=Evergreen.git;a=commit;h=679578d> |
11:46 |
pinesol_green |
[evergreen|Jason Stephenson] LP#1479953: Rename indexes to *_idx and add pgTAP test. - <http://git.evergreen-ils.org/?p=Evergreen.git;a=commit;h=57d756f> |
11:46 |
pinesol_green |
[evergreen|Jason Stephenson] LP#1479953: Stamping Upgrade Script - <http://git.evergreen-ils.org/?p=Evergreen.git;a=commit;h=768680b> |
11:47 |
bshum |
jeffdavis++ Dyrcona++ |
11:47 |
Dyrcona |
kmlussier++ # she tested it, too. |
11:48 |
bshum |
berick: I'm going to forward port your 2.8 translation changes to master so that the Czechs have something to start working with. |
11:48 |
berick |
bshum++ |
11:51 |
|
bmills joined #evergreen |
14:24 |
yboston |
OK |
14:25 |
yboston |
#action kmlussier will complete the marc stream importer work and check on the status of the RDA docs |
14:25 |
yboston |
#info 3) remingtron will send out an email to the Dig list anouncing that a new file extension will be used for AsciiDoc files |
14:25 |
remingtron |
I was made aware of a possible problem with the way git tracks history when a file name changes. So I want to test that before any more public announcements. |
14:25 |
yboston |
interesting |
14:26 |
yboston |
I understood that it usually catches name changes |
14:26 |
remingtron |
don't want to mess up git history too much, especially for a non-essential change |
14:30 |
Stompro |
I'll do it in a few minutes, no need for an action item... |
14:30 |
yboston |
anything else on this issue/topic before me move on? |
14:31 |
remingtron |
Stompro: you're that much more ahead on next meeting :) |
14:31 |
yboston |
moving on |
14:31 |
yboston |
#info 4) remingtron will begin testing use of and conversion to the .adoc extension |
14:32 |
yboston |
is there more to say about this? |
14:32 |
remingtron |
not yet, I'll keep working on it |
14:32 |
yboston |
no problem at all |
14:32 |
yboston |
remingtron++ |
15:28 |
csharp |
kmlussier: yboston learned that at Berklee College of Music :-) |
15:28 |
yboston |
1.25 |
15:28 |
kmlussier |
csharp++ |
15:29 |
csharp |
okay, so for the logs, I did a giant purge of old hold requests and that seems to have fixed the Clear Holds Shelf checkin modifier on our test server |
15:30 |
csharp |
now to figure out how to do the same on production without scheduling a several-hours-long downtime |
15:33 |
Dyrcona |
csharp: What did you do to purge the old holds? |
15:33 |
dbs |
yboston++ |
15:35 |
|
rlefaive joined #evergreen |
16:10 |
|
Dyrcona joined #evergreen |
16:34 |
berick |
haha, george_duimovich++ |
16:42 |
|
bmills joined #evergreen |
16:52 |
pinesol_green |
Incoming from qatests: Test Success - http://testing.evergreen-ils.org/~live/test.html <http://testing.evergreen-ils.org/~live/test.html> |
16:57 |
Dyrcona |
Does the live test do a fresh install and run tests? |
16:58 |
Dyrcona |
Does it run pgtap as well as perl tests? |
17:06 |
|
mmorgan left #evergreen |
18:07 |
|
bmills joined #evergreen |
18:10 |
|
jonadab_znc joined #evergreen |