|
we use hang fire to queue up tasks from our webapi app. the tasks are also c# but they're compiled into exe's and checked into our tree
|
# ? Dec 3, 2017 19:53 |
|
|
# ? Jun 8, 2024 17:49 |
|
the engineer who did it used the goon in a well development methodology
|
# ? Dec 3, 2017 19:56 |
tp question about updating sql tables. when i know what rows and fields i want to update, its p clear how and what. what if i have a table with dynamically, recursively updating values in previous columns? is there a better way to figure out the updates than to compare the updated table (entire table is returned by query) with the stored query result table - when it would make much more sense to just overwrite it?
|
|
# ? Dec 3, 2017 20:16 |
|
cinci zoo sniper posted:recursively updating values in previous columns? when you cut against the grain there isn't a good way to go about it
|
# ? Dec 3, 2017 20:35 |
tef posted:when you cut against the grain there isn't a good way to go about it got it, cheers. ill just overwrite the entire table then, its not much
|
|
# ? Dec 3, 2017 20:47 |
|
cinci zoo sniper posted:tp question about updating sql tables. As for when it would make sense to completely overwrite the table? Almost never.
|
# ? Dec 3, 2017 21:47 |
Lady Gagas Penis posted:I think we need more details on this one. im thinking now myself, if i really can't define a condition when a row will become immutable. ill need to check with coworker tomorrow, he devised this algorithm. if he wrote it through fixed starting point, the i guess i can just query for the incomplete rows, since those all are separate aggregation units, and then update them in a straightforward fashion (i intend storing aggregate of one table in a different table) now if the starting point is floating the it's balls. there's no definable end state, and data changes arbitrarily withing a very loosely defined timeframe. with the number of checks for changes than id need to do, overwriting seems like the only rational option
|
|
# ? Dec 3, 2017 21:57 |
|
cinci zoo sniper posted:tp question about updating sql tables. you mention "previous columns", but the relational algebra really has no concept of column or row ordering. so you're going to have a bad time trying to access a "previous column" when all the SQL verbs talk in terms of bags of objects with unordered properties. didn't you mention storing time series as rows in another post, with times as columns? is that really a thing people do? a performance optimization when you always want to query the whole series maybe? seems like it would be a pain to update.
|
# ? Dec 3, 2017 22:09 |
ShimaTetsuo posted:you mention "previous columns", but the relational algebra really has no concept of column or row ordering. so you're going to have a bad time trying to access a "previous column" when all the SQL verbs talk in terms of bags of objects with unordered properties. previous rows, my bad - even though it doesn't change much. and yeah, it's time series aggregates that in our case are helpful to monitor how well we are doing as a business
|
|
# ? Dec 4, 2017 05:10 |
can someone link me that series of blog entries about sql explain entries that got posted to yospos recently? at least i think they were, but i cant find them (not https://use-the-index-luke.com/) and the loving microsoft edge i used at work is not helpful in digging it up in history e: found it https://www.depesz.com/2013/04/16/explaining-the-unexplainable/ cinci zoo sniper fucked around with this message at 08:29 on Dec 4, 2017 |
|
# ? Dec 4, 2017 08:04 |
|
yeah i posted that a little while ago and i highly recommend it, the dude's english isn't 100% perfect but you can understand what he's saying and he goes into detail
|
# ? Dec 4, 2017 19:35 |
DELETE CASCADE posted:yeah i posted that a little while ago and i highly recommend it, the dude's english isn't 100% perfect but you can understand what he's saying and he goes into detail cheers! and don't worry about my familiarity with broken english
|
|
# ? Dec 4, 2017 19:42 |
|
regarding your update vs overwrite question, i'm not sure if the time series points matter here. you are going to modify a bunch of rows. if it's only a few then it will be faster to do individual updates. if it's every single row then it's faster to rewrite the whole table. the line between the two is not clear, i guess you'd have to experiment and benchmark. note that for large updates you may also be spending a lot of time updating indexes or checking constraints. in that case it is often faster to drop those, run the update, and recreate them again. thankfully postgres has transactional ddl
|
# ? Dec 4, 2017 19:42 |
|
also i'm guessing you already know this but given the amount of people in the history of my company (judging from its codebase) who did not... this is slow: code:
code:
just do this: code:
|
# ? Dec 4, 2017 19:53 |
|
Flat Daddy posted:the engineer who did it used the goon in a well development methodology i googled goon in a well and the first hit was a pastebin thing that described this perfectly
|
# ? Dec 4, 2017 20:13 |
i don't know anything, this literally is the first time i'm going beyond a select statement - if me being the most frequent poster itt hasn't convinced you already as for update vs overwrite, i had a brief meeting with the author of the scripts i'm storing, and there indeed is no way to logically describe the update condition. hence, delete table every time all the time. i'll still keep your advice though, in case ill foray further into non-select database things. i'd definitely do the second variant every time and just shrug were it to fail
|
|
# ? Dec 4, 2017 20:18 |
|
CRIP EATIN BREAD posted:i googled goon in a well and the first hit was a pastebin thing that described this perfectly Because I had to google it too: https://pastebin.com/F83ZG2L6
|
# ? Dec 4, 2017 20:59 |
|
gonadic io posted:Because I had to google it too: https://pastebin.com/F83ZG2L6
|
# ? Dec 4, 2017 21:01 |
|
i found myself reading stl header files today life tip: never do this
|
# ? Dec 5, 2017 00:36 |
|
st-hell, sounds like
|
# ? Dec 5, 2017 00:37 |
|
writing a file system in FUSE is fun got a nice Amazon S3-backed hybrid cloud file system going here with data deduplication. working on snapshots now
|
# ? Dec 5, 2017 02:43 |
|
well i did it, i wrote a signed division wrapper around the jag gpu's unsigned fixed-point division instruction and set up perspective divide of a triangle. it took me 2 evenings. i'm going to go do anything other than programming now
|
# ? Dec 5, 2017 03:25 |
|
play some factorio, imo.
|
# ? Dec 5, 2017 03:29 |
|
Shaggar posted:play some tis-100, imo.
|
# ? Dec 5, 2017 03:32 |
|
Soricidus posted:i found myself reading stl header files today clang has really nice STL headers and then you realize the template definition you're reading is actually a pleasantly formatted comment and the real defintion is a creeping horror later in the file
|
# ? Dec 5, 2017 03:38 |
|
okay, i lied, i had to just finish the feature i was working on https://twitter.com/LuigiThirty/status/937917925557665792
|
# ? Dec 5, 2017 06:42 |
|
Star War Sex Parrot posted:writing a file system in FUSE is fun is it immutable or do you have a separate metadata store
|
# ? Dec 5, 2017 06:45 |
|
Luigi Thirty posted:well i did it, i wrote a signed division wrapper around the jag gpu's unsigned fixed-point division instruction and set up perspective divide of a triangle. it took me 2 evenings. i'm going to go do anything other than programming now
|
# ? Dec 5, 2017 12:01 |
c tp s:
|
|
# ? Dec 5, 2017 12:05 |
|
What is c tp s
|
# ? Dec 5, 2017 13:01 |
|
current tp status
|
# ? Dec 5, 2017 13:03 |
TimWinter posted:What is c tp s c tp p s
|
|
# ? Dec 5, 2017 13:05 |
speaking of, im now trying to think, not too hard, about optimal indices for enum and char(n) columns
|
|
# ? Dec 5, 2017 13:06 |
|
TimWinter posted:What is c tp s Current Terrible Programmer(ing?) Status It's a riff on Current Job Status, I. E. The cjs thread
|
# ? Dec 5, 2017 13:49 |
|
cinci zoo sniper posted:speaking of, im now trying to think, not too hard, about optimal indices for enum and char(n) columns an index with worse (numerically higher) selectivity than 1% isn't really worth using not sure if that's the official term or metric for it but you get what i'm saying hopefully.
|
# ? Dec 5, 2017 15:30 |
Sapozhnik posted:an index with worse (numerically higher) selectivity than 1% isn't really worth using i didn't get anything. that means b-tree?
|
|
# ? Dec 5, 2017 17:00 |
|
If an index doesn't cut a query's search space down to 1% of the table's size or less then it isn't worth creating and maintaining. An enum doesn't sound like the sort of column that has a lot of distinct values relative to the size of the table, so the best kind of index to use on an enum column is no index at all. As with all things database though, it depends on the workload.
|
# ? Dec 5, 2017 17:10 |
Sapozhnik posted:If an index doesn't cut a query's search space down to 1% of the table's size or less then it isn't worth creating and maintaining. workloads are laughable so i wont bother then, cheers
|
|
# ? Dec 5, 2017 17:12 |
|
This story triggered my impostor syndrome pretty hard: https://marcan.st/2017/12/debugging-an-evil-go-runtime-bug/ The TLDR is: guy sees an intermittent segfault in a Go program, chases it down to a piece of bad RAM, then realizes that's not the core issue so he chases it down further to a kernel/compiler issue. But I felt the sheer breadth and depth of knowledge required to do all that was pretty impressive. I'd probably have welp'd out at "My Go program segfaults, file a bug".
|
# ? Dec 5, 2017 17:21 |
|
|
# ? Jun 8, 2024 17:49 |
|
minato posted:This story triggered my impostor syndrome pretty hard: https://marcan.st/2017/12/debugging-an-evil-go-runtime-bug/ if i had to choose between blaming code that i wrote and blaming a bug in, say, gcc, then yeah it's probably a bug in my code. if there's weird crashes in a go program on the other hand then yeah i'd totally be a bit more inclined to believe that rob "I Am Very Smart" pike hosed something up lol if you ever rely on go
|
# ? Dec 5, 2017 17:29 |