Time |
Nick |
Message |
08:15 |
|
sleary joined #evergreen |
11:15 |
|
sleary joined #evergreen |
13:32 |
csharp_ |
not expecting anyone to be around, but I'm experimenting with storing settings data as JSONB in postgres (in a non-production table on a non-production test server) because I'm interested in being able to edit the JSON without regexp_replace nonsense |
13:32 |
csharp_ |
wondering if eeevil or others have ever considered the benefits/drawbacks of just storing our JSON as JSON/B |
14:50 |
|
Dyrcona joined #evergreen |
14:55 |
jeff |
yeah, i wouldn't recommend regexp_replace to edit json, but I'm not sure I'd go so far as to use JSONB. |
14:56 |
Dyrcona |
I'd pull it out with Perl or Python and use a JSON library to modify it, probably. |
14:57 |
Dyrcona |
But I should check what the actual question is. |
14:57 |
* Dyrcona |
signed in to bug the folks over in #manjaro-arm. |
14:57 |
jeff |
JSONB has advantages (the joke is that the B stands for "better"), but it isn't json, and you can't round trip it. |
14:57 |
jeff |
you lose whitespace and ordering, for one thing. |
14:59 |
Dyrcona |
OK. Totally not what I thought the question was. |
14:59 |
Dyrcona |
My question is why JSON? Why not just tables with fields of the appropriate types? |
15:00 |
Dyrcona |
And, I don't usually think of sticking JSON in the database. I'd just use files in the filesystem. |
15:00 |
jeff |
I don't think that user settings or copy templates make sense to store as files in the filesystem. |
15:01 |
Dyrcona |
OK, but there's nothing in csharp_'s questions to suggest it is actually for Evergreen settings. |
15:02 |
jeff |
sure, I was making assumptions and applying external context to get to that conclusion. I could be very wrong. |
15:03 |
jeff |
csharp_: if this is about removing copy alert messages from templates, I'm somewhat more in favor of a different approach, but it might be unpopular. :-) |
15:04 |
Dyrcona |
I don't think I'm going to get an answer to my dependency issue in #manjaro-arm. I'm updating packages and two jdk packages contain the same files, so the package manager detects that and refuses to update anything. |
15:05 |
jeff |
roughly, having one copy alert that's maintained by trigger so that it appears in both asset.copy.alert_message and asset.copy_alert. Doesn't break reports then, either. :-) |
15:05 |
jeff |
Dyrcona: ah. ouch. |
15:06 |
Dyrcona |
Also, listening to radio from Den Haag, and I have to say, it rocks. |
15:13 |
jeff |
And I don't mean to dismiss JSONB -- it's really handy. I just didn't think wanting to manipulate JSON values in the database was a good reason to migrate from TEXT to JSONB. :-) |
15:14 |
jeff |
there probably isn't anything client-side that would care about the whitespace changes, but I suspect a few places (or humans) DO care about loss of ordering, so we'd probably need to inject a sort order into the JSON if we did that. |
15:15 |
Dyrcona |
I don't think should be manipulated much in the database, hence my previous suggestion of manipulating it in the program. The corollary is use the database as the backing store. |
15:16 |
Dyrcona |
The order shouldn't matter, but I can imagine a poorly implemented JSON library that could care. |
15:17 |
Dyrcona |
The "answer" to my issue is bug #archlinuxarm if they're responsible for the packages, otherwise ask on the forum. |
15:19 |
Dyrcona |
Duth radio playing Dropkick Murphys... :) |
15:19 |
Dyrcona |
s/Duth/Dutch/ |
15:20 |
Dyrcona |
I should stop using derived distros. Maybe it's time I take Linux from scratch seriously. |
17:04 |
pinesol |
News from commits: LP#2022366: quell a GitHub actions warning <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=b45b8ea4d7d50168908d142cb75fafed9adfb054> |
17:49 |
csharp_ |
jeff: indeed I'm thinking of alert_message in the cat.copy.templates - I guess an alternative is pull out the value, use perl or whatever to remove things, then shove it back in? |
17:50 |
csharp_ |
(thanks for responding with nuance on a Saturday too :-) ) |
19:42 |
|
sleary joined #evergreen |