|
speaking of, groovy log day 1: so far this feels like someone looked at java code and went" you know what this really needs is to have syntax more like python"
|
# ? Jan 22, 2018 17:51 |
|
|
# ? Jun 8, 2024 13:54 |
|
a lot of bad java libraries are also written by people who seem to want to write python code in java, the general attitude being "type systems are hard, let's do duck-typing by reflectively invoking everything"
|
# ? Jan 22, 2018 18:06 |
|
Sapozhnik posted:a lot of bad java libraries are also written by people who seem to want to write python code in java, the general attitude being "type systems are hard, let's do duck-typing by reflectively invoking everything" i for one really appreciate type errors getting deferred to runtime!!! big improvement a+++++
|
# ? Jan 22, 2018 18:10 |
|
Sapozhnik posted:a lot of bad java libraries are also written by people who seem to want to write python code in java, the general attitude being "type systems are hard, let's do duck-typing by reflectively invoking everything" just add this annotation to your class and everything will work for you automatically! sure it adds a ton of overhead but gently caress it, our 30 algorithm needs 4000 lines of reflection inspection code
|
# ? Jan 22, 2018 18:11 |
|
cinci zoo sniper posted:im not desperate. scala is just a logical technical extension of my current career i love scala so imo go for it
|
# ? Jan 22, 2018 18:15 |
|
HoboMan posted:speaking of, groovy log day 1: funny thing -- that was the entirety of the design process
|
# ? Jan 22, 2018 18:15 |
|
gonadic io posted:Yeah scala is your best bet here, it is the jvm analogue to f#. Different language design, much less focused/strict but it's still the functional jvm lang. Kotlin kinda looks interesting too in the space but it's not as mature as scala. Probably much better designed though. scala and f# are both functional but they have radically different philosophies of language design scala is "big" language, in the sense that c++ is a big language. they jammed everything in there. every feature that was even vaguely conceivable on the jvm is mashed into scala's design, somewhere. as with c++, you should not try to use all the features.
|
# ? Jan 22, 2018 18:17 |
|
fyi if you haven't looked into it, kotlin is a really good jvm language with top-notch java interop. its a really cool language and i hope it continues to take off
|
# ? Jan 22, 2018 18:18 |
|
code:
? ???
|
# ? Jan 22, 2018 18:24 |
|
HoboMan posted:
language features like this are so loving stupid and the only places i ever see them used is in really bad phd code. nothing says "i have a phd and my entire career has been in acadademia" like x1, x2, i1, i2, y3, z5 = doAthing() andDoThis() + zz1
|
# ? Jan 22, 2018 18:30 |
|
"what is variable scope?" - a person with a phd in computer science
|
# ? Jan 22, 2018 18:31 |
|
so can this be sanely written asJava code:
|
# ? Jan 22, 2018 18:40 |
|
cinci zoo sniper posted:same, i had to generate ssh key for them when they joined, much like painstakingly explain basics of git and gitlab, what cron is, and a number of other things any server-side focused it professional should be ashamed to have never heard about in 2018. also, a torrent of clueless questions like "cant we disable this ssh thing" or "can i check if vcs refresh operation in this cron worked via ftp" or "are all our servers linux? why dont we use windows? could i get a windows server?" you should really really try hard to see if you can convince someone with the authority to fire them, if you could fire them
|
# ? Jan 23, 2018 00:10 |
|
Convoolio posted:Trip report: agree but curious how/why you’re using helm
|
# ? Jan 23, 2018 00:12 |
|
how much of ctci does a terrible, useless, gormless idiot need to be able to get through to get jobs
|
# ? Jan 23, 2018 01:40 |
|
Convoolio posted:Trip report: except for helm, its bad
|
# ? Jan 23, 2018 04:05 |
bob dobbs is dead posted:you should really really try hard to see if you can convince someone with the authority to fire them, if you could fire them that's 100% impossible, and maybe not even necessary, but ill need to be taking up with boss that there are decision he should be let to make, and that normal functioning.will require a dedicated person for tech support initially, i can't pull half-time it support duties on top of my already slightly overloaded schedule as is
|
|
# ? Jan 23, 2018 06:03 |
|
cinci zoo sniper posted:that's 100% impossible, and maybe not even necessary, but ill need to be taking up with boss that there are decision he should be let to make, and that normal functioning.will require a dedicated person for tech support initially, i can't pull half-time it support duties on top of my already slightly overloaded schedule as is you’re running rapidly towards a tech lead title or equivalent, be sure to get the figgies to go with it
|
# ? Jan 23, 2018 06:40 |
eschaton posted:you’re running rapidly towards a tech lead title or equivalent, be sure to get the figgies to go with it i got a "lead" title recently, but will give it until summer to.change figgies or change employer. preferably the former, since im doing well but lack much of a paper trail to convince a random recruiter to.give me a shot
|
|
# ? Jan 23, 2018 06:46 |
|
AWWNAW posted:agree but curious how/why you’re using helm managing differences in dev/production configs, programmatically generating container specs, generating lists of environment variables within the containers it takes a bit of thinking but I don't know of anything else that can do the same, as well
|
# ? Jan 23, 2018 09:28 |
|
It's not better in EVERY way and this is a generalization, but if you're a competent but green person then landing somewhere with a bunch of incompetent people is the dream first gig.
|
# ? Jan 23, 2018 15:17 |
pangstrom posted:It's not better in EVERY way and this is a generalization, but if you're a competent but green person then landing somewhere with a bunch of incompetent people is the dream first gig. while i dont think im all that competent either, its definitely an interesting gig, and looks nice on paper and in bank account. my major dislikes are hectic/unstructured everything and lack of people to discuss with or learn from the technical stuf other than sql (or php, but its not relevant to my job)
|
|
# ? Jan 23, 2018 15:27 |
|
You are definitely more competent than most but agree, that's the main downside. But you get a chance to own/build/change things yourself and to skip a paygrade or three if you can stay on decent terms with people. Also, I'm only (incompetent) developer-adjacent, this is just something I've seen happen a few times and maybe it's not representative.
|
# ? Jan 23, 2018 15:46 |
pangstrom posted:You are definitely more competent than most but agree, that's the main downside. But you get a chance to own/build/change things yourself and to skip a paygrade or three if you can stay on decent terms with people. Also, I'm only (incompetent) developer-adjacent, this is just something I've seen happen a few times and maybe it's not representative. not sure its representative, but thats more or less what happened. i skipped 3 titles and two paygrades - will be negotiating third one to bridge it to the latvian "6 figgies" equivalent after i finally get a major project done. with our "startup style" everything ive managed to do basically nothing related to my primary job description in the 6 months that i've worked here. but yeah freedom to define things however and do whatever is really nice, im not sure i comprehend well how i would feel/do in an environment where im just told "here, hack this like that"
|
|
# ? Jan 23, 2018 16:43 |
|
pangstrom posted:It's not better in EVERY way and this is a generalization, but if you're a competent but green person then landing somewhere with a bunch of incompetent people is the dream first gig. https://thedailywtf.com/articles/The-Difference-Between-Better-and-Less-Bad
|
# ? Jan 23, 2018 17:41 |
|
The awful thing about getting it right the first time is that nobody realizes how hard it was.
|
# ? Jan 23, 2018 17:43 |
somewhat relatable yea. for an example from my job, before our local team was hired my employer was unable to identify which services our customers paid for. to do that, you would have to do an sql query on a number of tables all together storing the information on payments, and then filter it on an unindexed datetime field to match against a separate set of tables with unindexed datetimes that store the information on services provided. you ask what happens if a customer ordered two services with a single order, or two services relatively close enough? tough poo poo sherlock, they have a generous window to pay down their outstanding balance after the fact of order. what about 3 50 dollar invoices and a 127 dollar payment? let's ask our resident astrologist
|
|
# ? Jan 23, 2018 17:49 |
have i mentioned tables with no primary keys or tables where field names are reserved keywords?
|
|
# ? Jan 23, 2018 17:51 |
|
people at my company like using Postgres like mongo. filled with lovely JSON currently working on a query with multiple joins against keys in JSONB fields some of these people have decades of experience!
|
# ? Jan 23, 2018 18:04 |
AWWNAW posted:people at my company like using Postgres like mongo. filled with lovely JSON we kidna do this too, but in json not jsonb, and it's an external data vendor stuff. there are some inhouse jsons, but they are tiny on average and barely used for anything. optimally would have it all at least in jsonb i guess, but so far im just happy it is stored as is, and not in mongo. gently caress, im starting to remember our xml storage in mongo
|
|
# ? Jan 23, 2018 18:09 |
c tp s: one of our business divisions has a new manager who, on their first they of office, did propose that we discard raw data and store only parsed inputs they have yet to even talk to anyone here
|
|
# ? Jan 23, 2018 18:12 |
|
AWWNAW posted:people at my company like using Postgres like mongo. filled with lovely JSON currently working towards this model because we're on mongo and desperately want to get away and the only way to do it iteratively is to start dumping poo poo into jsonb columns and gradually expand into a normalized database. there's just no other way around it. but lol i just wrote all the library code to support it i'm not gonna do the actual lovely work because i put in my notice.
|
# ? Jan 23, 2018 18:22 |
|
MALE SHOEGAZE posted:currently working towards this model because we're on mongo and desperately want to get away and the only way to do it iteratively is to start dumping poo poo into jsonb columns and gradually expand into a normalized database. congrats
|
# ? Jan 23, 2018 18:36 |
|
Sapozhnik posted:congrats thanks. by all accounts, new place has their head screwed on straight. i've really really enjoyed talking to the CTO/people I interviewed with, and I have a friend who I trust that works there currently and says lots of good things. cons: One of the things i like about my current job is the people. I'm actually going to miss socializing with coworkers. that feels like a big loss, especially for an anti social nerd.
|
# ? Jan 23, 2018 18:38 |
MALE SHOEGAZE posted:currently working towards this model because we're on mongo and desperately want to get away and the only way to do it iteratively is to start dumping poo poo into jsonb columns and gradually expand into a normalized database. congrats, but also what would be your long term arch in that case? raw data in some corner and then parse it out into more and more branching structure in separate tables?
|
|
# ? Jan 23, 2018 18:40 |
|
cinci zoo sniper posted:have i mentioned tables with no primary keys or tables where field names are reserved keywords? reserved keywords are annoying but shouldn't be a real problem. if you forget your square brackets you'll just get a syntax error, and it's easier to always use the brackets everywhere no pkeys on the other side is a true horror, but you should be able to fix it in like half an hour
|
# ? Jan 23, 2018 18:57 |
NihilCredo posted:reserved keywords are annoying but shouldn't be a real problem. if you forget your square brackets you'll just get a syntax error, and it's easier to always use the brackets everywhere reserved keywords are not a problem for "normal" sql queries, but our recently hired etl dude uses some pentaho tool afaik and he's been cursing about that a plenty so maybe it has some issues there or something? as for pkeys, im not sure what would a fix entail. i have yet to knowingly come across one of those tables, thankfully. our usual catch22 for this kind of issues like no pkey or double indices or no relevant indices or other stuff like that is lack of dedicated database people. so i come up and say like "yo we have this 1 billion row table that is not partitioned and doesnt have datetime indexes, and we need to filter by datetime column" and whichever web developer the it project managers did assign this to just kind-of shrugs and says "is big table. no do. other option?"
|
|
# ? Jan 23, 2018 19:07 |
|
cinci zoo sniper posted:reserved keywords are not a problem for "normal" sql queries, but our recently hired etl dude uses some pentaho tool afaik and he's been cursing about that a plenty so maybe it has some issues there or something? i've never encountered a serious tool that autogenerated sql and didn't properly escape object names. pentaho is supposed to be a serious tool, so if it chokes on a poorly-chosen but legal column name i'd straight up open a ticket. quote:as for pkeys, im not sure what would a fix entail 1) alter table butts add primary key (buttID) 2) did it work? if yes, great, you're done. if no, you need to extract the unique values 3) select buttId, max(buttSize) into butts_unique from butts group by buttId. replace max() with whatever grouping function is more appropriate for each column 4) alter table butts_unique add primary key (buttId) 5) drop table butts. alter table butts_unique rename to butts. (you may have to drop and recreate foreign keys, indexes etc. as well, but if you lack primary keys there won't be many of those too)
|
# ? Jan 23, 2018 19:24 |
|
this is mssql, right? if you have a time series table then make a clustered index on the timestamp column. what are the access patterns like though? does that row and its timestamp get updated at all? do you need to audit modifications? if you have a literally 1 billion row table and you need to be able to select ranges of timestamps out of it then that table definitely needs to be partitioned, though managing partitioned tables is never anything less than a complete management nightmare so hf with that i guess. need to add a primary key? alter table crazybullshit add column id uniqueidentifier primary key default newid() not null;
|
# ? Jan 23, 2018 19:26 |
|
|
# ? Jun 8, 2024 13:54 |
|
use newsequentialid or w/e
|
# ? Jan 23, 2018 19:28 |