|
cool av posted:how do ppl fee about http clients throwing exceptions on non-2xx responses? I’m leaning toward I don’t like it. i disapprove too exceptions work best for errors that can't be pinned down to a specific function call - out of memory (including stack overflows), out of storage, lost permissions, lost connection to the db... nasty failure modes, where the least bad option is a high-level catch statement that aborts the whole operation and maybe goes for a retry if the error is very clearly and very predictably tied to a single line of code, like httpclient.postAsync() or json.deserialize(), then returning an error value is pretty much strictly better. using try/catch around a single line of code is analogous to using an if/switch/match on the result value, except uglier, slower, and more error prone
|
#
¿
Nov 17, 2023 18:03
|
|
|
|
| # ¿ Oct 4, 2024 13:04 |
|
|
Cabbage Disrespect posted:what's that, you say? non-deterministic classpath issues? sometimes a method wants a javax.ws.Thing but we feed it jakarta.ws.Thing and sometimes vice-versa? very cool. here's a nickel dude, go buy yourself a quay.io/keycloak/keycloak:23.0
|
|
#
¿
Dec 7, 2023 21:55
|
|
|
redleader posted:x postin because im still mad about this. the .net date/time types are absolute trash and they should just do what java did and steal the good date/time library outright system.datetime is only suitable for poo poo code that never leaves your timezone and should come with a compiler warning for using it at all. it's on the same level as hardcoding user-facing strings or writing non-responsive UIs. if you get a datetime from a library, *immediately* convert it to a datetimeoffset system.datetimeoffset is decent i'll still install nodatime on any project that i actually give a poo poo about, but now that they've finally added DateOnly and TimeOnly types it's less mandatory
|
|
#
¿
Dec 11, 2023 11:27
|
|
|
redleader posted:row polymorphism is at least a decade away from hitting any sort of normie- or normie-adjacent lang. you can quickly tell because a quick goog search only returns a bunch of comp sci theory blogs and papers typescript literally has row polymorphism right now dude JavaScript code:omit<T, "list of destructured fields"> is the statically-checked type of 'rest'
|
|
#
¿
Jan 2, 2024 13:05
|
|
|
VikingofRock posted:Is the "...rest" here needed as part of the definition of row polymorphism? I'm asking because I'm wondering if C++ would count as row polymorphic with something like this: i think the copy constructor is performing the same role as "rest" here, that is preserving the fields you don't care about unaltered into the return value - which is the bare minimum to be considered RP (in more high-level languages you do RP with records so the constructor is implied by that restriction) though, RP can also include the more powerful ability to add/drop fields from a record while preserving the others. purescript can do it (i think this is where I first encountered it, and i thought it was a requirement to be RP), and now typescript can too i imagine c++ might be able to do it with a lot of template fuckery - you need to encode a relationship between the constructors of different types.
|
|
#
¿
Jan 3, 2024 09:56
|
|
|
Subjunctive posted:write an `equals_ignoring_timestamp` method and move on with your life the annoying issue, mr. president, is that when someone adds another field to the struct, they will get exactly zero warnings from the compiler that somewhere in the codebase there is an "equals_ignoring_timestamp" method that compares 9/10 fields and now you just made it 9/11 whoops
|
|
#
¿
Jan 11, 2024 11:36
|
|
|
Subjunctive posted:that’s OK, the tests will break mein führer...
|
|
#
¿
Jan 11, 2024 19:50
|
|
|
I've never rolled back in prod either, and if I had to it would definitely warrant a postmortem but knowing that I *could* gave me that little extra feeling of safety, so imo it was well worth the three minutes it took to write an ALTER TABLE or two for the rollback statement
|
|
#
¿
Jan 15, 2024 20:58
|
|
|
my homie dhall posted:if you hate migrations so much seems like you could create a new table for each version of your schema? I work on fake things which are allowed to have downtime, so I just do migrations even better, make sure your schema is in sixth normal form and your migrations well never need locks
|
|
#
¿
Jan 16, 2024 15:43
|
|
|
advertising is negative-sum economic activity bitcoin may cause the extinction of sixteen species to transfer one dollar, but at least a money transfer and account balance is a useful thing that people want advertising is literally paying to make other people unhappy
|
|
#
¿
Jan 16, 2024 17:30
|
|
|
mystes posted:Handling resizing in vb6 sucked because IIRC you had to handle it manually in code, whereas in modern winforms there are actual layout controls that will handle it for you which makes it a million times easier. had to handle resizing in a winforms app, vbnet not even vb6, can confirm it sucked balls you had a few basic anchoring flags but otherwise it was lots and lots of code in the Resize handler where you had to do manual pixel math
|
|
#
¿
Feb 20, 2024 17:54
|
|
|
ChickenWing posted:jesus christ how horrifying we eventually got to something pretty similar with emscripten and later wasm "i have a bunch of code that needs to be written in a Serious Language(tm) but must also run on the most universal client available" isn't a problem that's going to go away
|
|
#
¿
Feb 22, 2024 17:16
|
|
|
finally, i can run clojure in the browser
|
|
#
¿
Feb 22, 2024 22:38
|
|
|
Sagacity posted:Pro-tip (coming from a bunch of experience, ymmv): Consider how much time you'd have to spend on writing the transforms manually. I bet it is going to be less than the time you'd be spending configuring tools to 'magically' transform objects and keeping those dependencies updated. This. Also, if your language is strongly typed and all constructor parameters are mandatory, you can let Copilot write the basic "x.Foo = y.Foo" stuff fairly safely. Then you'll have a bunch of boring, repetitive, but valuable code - valuable because it will break with a compiler error when there's a change on either side of the data flow that you need to consider. It's much better than adding a new field to the domain object and have that magically and silently appear in the DTO even though the service on the other side has not been updated to handle it! Also, code organisation suggestion: there is only one pure domain object for business logic, but potentially many DTOs for different targets. So don't add "toSqlDto" and "fromSqlDto" methods to your domain object, instead add "toDomain" and "fromDomain" methods to each of your DTOs. Keep the dependencies one-way.
|
|
#
¿
Feb 28, 2024 10:31
|
|
|
Its a Rolex posted:the project i work on is built around 365 days in a year (or more specifically, 8760 hours in a year). the idea of a leap day breaks the entire thing at a very fundamental level. 8760 was a magic number sprinkled everywhere in the code when i started, and now it's just burned into my brain as somewhat-useful trivia curse those obscure edge cases! who could possibly have foreseen them
|
|
#
¿
Mar 1, 2024 10:02
|
|
|
necrotic posted:I think this is a neat solve and also a very glad it’s not my problem. im very glad we've found a way to deal with leap seconds and our time libraries can be stable for the foreseeable future On 18 November 2022, the General Conference on Weights and Measures (CGPM) resolved to eliminate leap seconds by or before 2035. The difference between atomic and astronomical time will be allowed to grow to a larger value yet to be determined. A suggested possible future measure would be to let the discrepancy increase to a full minute, which would take 50 to 100 years, and then have the last minute of the day taking two minutes in a "kind of smear" with no discontinuity.
|
|
#
¿
Mar 4, 2024 11:43
|
|
|
Antigravitas posted:I normally write in Python 3, and we are pretty good at moving to new versions. I know you may not have full control of the toolchain, but isn't this exactly what Babel was made for? e: babel won't give you a debugger but if you are able to write in typescript that takes care of most of those dumb "object does not have this particular property/method" time-wasting errors
|
|
#
¿
Mar 11, 2024 19:09
|
|
|
Private Speech posted:I was reading about planning poker since it came up itt recently and I really enjoyed this cop out bit is that specific quote about planning poker? because "veto power" only makes sense in a yes/no vote
|
|
#
¿
Mar 20, 2024 16:36
|
|
|
Subjunctive posted:veto doesn’t need to have anything to do with a vote, it’s just the ability to prevent an action, like “making 5 days the recorded estimate” ok then what happens after they use the veto? if the estimate goes back to the devs, it's a useless veto (the devs have no reason to back down to break the stalemate, it's the managers who need to deliver) if it goes to managers, it's not a veto and it contradicts the previous sentence (that managers don't vote)
|
|
#
¿
Mar 20, 2024 16:57
|
|
|
Deep Dish Fuckfest posted:just looked at the jpeg xl github and i hate their logo and i hate the whole thing on that basis alone. also the logo is in svg; way to have faith in your own format from your description i was expecting some imagemagick-style "graphic design is my passion" embarrassment. it's just extremely basic and i do not understand how it could have made you so mad
|
|
#
¿
Apr 5, 2024 18:35
|
|
|
the XML serializer in c# does have a magic attribute that is supposed to mean "when deserializing, check the value of this field to decide which subclass to target" i never got it to work and just wrote a switch instead
|
|
#
¿
Jun 8, 2024 11:10
|
|
|
toiletbrush posted:Its still 1000x less evil than the repo at current job where they decided to use C#'s implicit operator to do all the mapping nah, not really. an implicit operator is still type-safe and debuggable
|
|
#
¿
Jun 9, 2024 10:17
|
|
|
Bloody posted:I just use fork and it’s so good at git that I haven’t needed the terminal in years. 
|
|
#
¿
Jun 13, 2024 09:38
|
|
|
Corla Plankun posted:people who can't figure out/tolerate cli git should be more embarrassed imo ah yes, the cli interface that has stood as the arch-example of inconsistent design and crappy ux for two decades people should definitely make a point of pride of using that. perhaps also relying on nothing but their own wits and the official man pages, which have also been a long-running joke of awkwardness
|
|
#
¿
Jun 14, 2024 03:53
|
|
|
I had the displeasure of using TFS and it had "shelving". Jetbrains has been around since 2000, perhaps all those weird terms predate git and when they added git support they decided to keep the pre-existing names for the features supported by multiple VCSes?
|
|
#
¿
Jun 14, 2024 20:43
|
|
|
dilbert is actually very proud of his strobe tie, thank you
|
|
#
¿
Jun 21, 2024 10:42
|
|
|
Ultrapotassium posted:What's the alternative here, throwing an exception on the child thread? In really old c#, before Tasks let alone async/await, this was not uncommon. You'd fire off your async job by starting a full fat System.Threading.Thread, at which point it was very tempting to just call Abort() whenever the user clicked Cancel or such. Even in the One Direction years this was known as very bad practice. The MSDN article on Thread.Abort() is a giant list of ways it can gently caress up, and they finally killed the method for good in .NET 5 (something something abortion joke)
|
|
#
¿
Aug 8, 2024 16:57
|
|
|
bob dobbs is dead posted:too many peeps know the fate of griswold and volrath cast iron pans for that to happen, i think ironically that *is* something that a subscription model solves, however loathsome it may be when I recently moved to a new city I got a swapfiets subscription bike for a while. guess what? it was built like a loving tank, because maintenance was a cost and not a revenue source for the vendor
|
|
#
¿
Aug 24, 2024 03:24
|
|
|
i naively thought certification meant "you can sue us for damages if there's an undocumented bug (and it did enough damage to be worth litigating)" is it actually just vibes?
|
|
#
¿
Aug 26, 2024 07:40
|
|
|
Nomnom Cookie posted:1k repos 
|
|
#
¿
Aug 30, 2024 06:28
|
|
|
Bourricot posted:I feel seen. on my previous project, we had 66 repos for a team of 5 devs (the company standard was 2 repos: 1 frontend, 1 backend) not quite that bad but at my old job, when I got to lead my first product, I did make three repos for backend, web frontend, and android app in my naive youth I believed the ceo when he said that we'd be open to third-party clients, so I wanted teams to treat the openapi contract as ironclad and carefully versioned, while still being able to work at their own paces given the different team sizes of course nobody ever cared about making a client and the team never grew big enough that they wouldn't fit into a large room, so it was 100% pointless overhead. but even if that hadn't been the case, the CI check for changes in the generated openapi.json was what actually helped to limit accidental breaking changes, and it would have worked just as well in a monorepo
|
|
#
¿
Aug 31, 2024 10:03
|
|
(so the tech lead can pad his resume)
|
Powerful Two-Hander posted:it's pretty common for corps to try and put a "code escrow" clause in with the theory being that if you go under they could somehow continue to use the product by building and deploying from scratch. I always used to say "lol, lmao" to this and be like "guys we can barely build our own code this is insane" but they'd insist and the vendor would just shrug and go "ok that's another 10k though" and ofc now everything is deployed to 900 billion different Aws components it's even less likely that it would be possible to ever do this so I just tell them to drop the clause entirely. for an opposing view - and I must disclaim that my experience is with European SMEs rather than megacorps - code escrow is something I actively pushed for as a developer because I believe it's one of the very rare instances where corporate interests and good SWE practices can align, so it should be encouraged as much as possible typically nobody gives a drat how annoying or brittle your build is as long as they get the shipped artifact, and when you log hours for ci/cd work or documentation they act like you made that poo poo up. so having an actual contractual term that justifies that expense is a good thing
|
|
#
¿
Sep 20, 2024 16:05
|
|
|
we started being stricter with the hell customer and turning every one of their "just make it X" tickets into accurate specs with realistic estimates through painstaking q&a sessions (which frequently find out that the request didn't make any sense in the first place) now they say they're unhappy about the hours it takes to figure out the exact specs and estimates. the compromise our manager reached is that for every off-the-cuff "make it X" ticket we will give an equally off-the-cuff "it could take 2 days or a week, maybe" estimate, and the hell customer will decide if they want to put in the time to actually figure out what they want 
|
|
#
¿
Sep 23, 2024 14:13
|
|
|
when i encountered that style of writing in the wild for the first time - blissfully unaware of its origin - i remember thinking, how much of a pompous blowhard do you have to be to preface 'as a developer...' to every one of your mundane-rear end tickets asking for a new command-line flag
|
|
#
¿
Sep 23, 2024 21:45
|
|
|
|
| # ¿ Oct 4, 2024 13:04 |
|
|
raminasi posted:where this gets wrinkly is when the thing you're trying to figure out is how to surface errors to the end user in the first place, and they have no idea why you want to do that, because no errors should ever be possible. why are you writing the application so it can have errors? aren't you good at your job? tell them a cosmic ray flipped the sign in the field
|
|
#
¿
Sep 26, 2024 13:30
|
|