|
Shaggar posted:i've had to explain .net user secrets to the same people over and over. i stick a readme in the solution with the structure of the user secrets so all they need to do is right click the project -> manage user secrets -> paste in the template -> fill in sensitive details. i make users fill in environment variables everytime, and have startup.cs set to fail if one is unset.
|
# ? Apr 27, 2022 15:04 |
|
|
# ? Jun 8, 2024 06:45 |
|
i should probably add something to validate the settings when in dev, but i think i could throw an exception that tells them their user secrets arent set as specified in the readme and they'd still ask me whats wrong. i like the user secrets vs environmental variables cause its makes per-project settings very easy and they can be structured like normal appsettings.
|
# ? Apr 27, 2022 15:14 |
|
i hate using appsettings too lol, i try to avoid having any files my build depends on that arent code
|
# ? Apr 27, 2022 15:18 |
|
the kind of person who won't do something as simple as set a user secret sure as poo poo isn't going to read the message in an exception
|
# ? Apr 27, 2022 15:18 |
|
but im a fuckin idiot
|
# ? Apr 27, 2022 15:19 |
|
I've started going over pull requests I've submitted comment by comment, tallying which I could have anticipated and avoided beforehand, because the question "wait, am I really this stupid?" has sustained itself long enough that I need to collect data to regain my sanity and certainty.
|
# ? Apr 27, 2022 17:17 |
|
RokosCockatrice posted:I've started going over pull requests I've submitted comment by comment, tallying which I could have anticipated and avoided beforehand, because the question "wait, am I really this stupid?" has sustained itself long enough that I need to collect data to regain my sanity and certainty. The answer to your question is yes. But your coworkers likely aren't any better.
|
# ? Apr 27, 2022 18:18 |
|
gently caress you, a database call should not RTE in a way that determines the HTTP response code.
|
# ? Apr 27, 2022 18:21 |
|
here is your extremely cursed article title for today: https://www.collabora.com/news-and-blog/news-and-events/socketcan-x-kubernetes.html
|
# ? Apr 27, 2022 22:43 |
|
Sapozhnik posted:here is your extremely cursed article title for today: quote:One component of our workflow required the availability of a CAN interface inside the Kubernetes Pod; to our surprise, such support didn't exist gonna be honest that didn't surprise me very much
|
# ? May 3, 2022 16:20 |
|
lmao CAN on k8s
|
# ? May 3, 2022 16:31 |
|
CRIP EATIN BREAD posted:lmao CAN on k8s CANBUS to k8s interface can’t wait to solve that one at Tesla
|
# ? May 3, 2022 16:53 |
|
*nervously trying to remember the exact kubectl command to open the emergency exit doors that i need to type on the touch screen command line that comes up after entering the konami code as the encroaching flames from the battery fire start to melt the plastic upholstery*
|
# ? May 3, 2022 17:21 |
|
I'm now on an "architecture standards review" committee and one thing up for agreement is 'react plus typescript for UI" so I went and looked up some react tutorial and lol, this is what people thing is easy and good? god help us
|
# ? May 13, 2022 21:57 |
|
love to have a dom implementation entirely in javascript to get around browser limitations
|
# ? May 13, 2022 21:58 |
|
Powerful Two-Hander posted:I'm now on an "architecture standards review" committee and one thing up for agreement is 'react plus typescript for UI" so I went and looked up some react tutorial and lol, this is what people thing is easy and good? god help us Could be worse... could be angular.
|
# ? May 13, 2022 21:59 |
|
js land hasn't found anything better than react in the last seven years or so the new react stuff (hooks and such) is a bit weird though and i avoid it for my own personal stuff but that is a somewhat controversial position
|
# ? May 13, 2022 22:11 |
|
I guess I can see the benefit of the event auto wireup or whatever, because with asp mvc it is kind of a pain to juggle model identifiers to append to lists or whatever, but still, at least your UI and DOM are separate from your controller and model code (don't ask about the grid that renders coloured cell content using strings containing html that are generated as varchar output from the database)
|
# ? May 13, 2022 22:22 |
|
frontend is a wasteland
|
# ? May 13, 2022 22:29 |
|
DELETE CASCADE posted:frontend is a wasteland backend too
|
# ? May 13, 2022 22:30 |
|
hoo boy wait until you hear about games
|
# ? May 13, 2022 22:31 |
|
at least webshit and enterpriseland are extremely extremely cushy wastelands with figgies
|
# ? May 13, 2022 22:38 |
|
Game developers have an excuse, but they still manage to render really complex scenes in 16ms or less. Web devs need 10 seconds and 10MB of network traffic to render a some text and images floating in a white void.
|
# ? May 13, 2022 22:45 |
|
Powerful Two-Hander posted:I guess I can see the benefit of the event auto wireup or whatever, because with asp mvc it is kind of a pain to juggle model identifiers to append to lists or whatever, but still, at least your UI and DOM are separate from your controller and model code (don't ask about the grid that renders coloured cell content using strings containing html that are generated as varchar output from the database) Time for the latest iteration of my anti-MVC rant. MVC is a fundamentally broken design and I don't know why it isn't completely discredited at this point. The "model" never serves to abstract the view from the controller or vice versa, they are both intimately aware of each other and changes to one part of the MVC trio inevitably require a change to all three. Also view templating technologies are without exception godawful. Firstly because the great majority of them work by bashing together strings with no regard for the syntax of HTML, so you get to deal with either visible markup or XSS exploits all over the place. Secondly because they are crappy castrated Turing-complete languages with abysmal tooling when you already have a perfectly good Turing-complete language that you are using to write the rest of your application, but that main Turing-complete language lacks convenient features to work with HTML elements as a first-class data type. React is a post-MVC user interface library. It gives you a syntax for working with HTML literals as a first-class data type and doesn't attempt to impose a contrived abstraction where none exists. There's also the desired-state aspect of the HTML that it emits but that's an entirely separate topic (but this aspect is also good). Yes the syntax is a bit weird. So what. If weird syntax is the biggest problem I'm having while writing a program then I am having a very good day all things considered.
|
# ? May 13, 2022 22:50 |
|
I just wrote some C++ template bullshit that ends with four >s. I don't know if that means I'm getting better at it or worse.
|
# ? May 13, 2022 22:59 |
|
Powerful Two-Hander posted:I'm now on an "architecture standards review" committee and one thing up for agreement is 'react plus typescript for UI" so I went and looked up some react tutorial and lol, this is what people thing is easy and good? god help us i hate all frontend tech but typescript and react has pretty good foundations, design principles, etc. and it feels intuitive at times.
|
# ? May 13, 2022 23:19 |
|
Sapozhnik posted:js land hasn't found anything better than react in the last seven years or so I also have misgivings about React Hooks. Their ergonomics are really strange and I can't say I get how they're meant to be a significant improvement.
|
# ? May 13, 2022 23:24 |
|
i love react hooks personally, i find them super ergonomic when you compose them
|
# ? May 13, 2022 23:34 |
|
Powerful Two-Hander posted:I'm now on an "architecture standards review" committee and one thing up for agreement is 'react plus typescript for UI" so I went and looked up some react tutorial and lol, this is what people thing is easy and good? god help us Hooks add a lot of complexity especially because they have a bunch of requirements that can't actually be enforced by the language even though I guess they're supposed to be simpler than actual FRP. It's also annoying that there are now so many different overly complicated libraries for handling state in react. Also I feel like I've posted this a zillion times but a vdom is basically a requirement if you want something approximating an elm style approach (which react still is even though it's wackier with hooks) because you just get a new view as the output (rather than a list of instructions on how to modify the DOM or something like that) and there's no way to figure out how to modify the old view to obtain the new view without diffing. mystes fucked around with this message at 23:44 on May 13, 2022 |
# ? May 13, 2022 23:40 |
|
Doom Mathematic posted:I also have misgivings about React Hooks. Their ergonomics are really strange and I can't say I get how they're meant to be a significant improvement. imo the biggest benefit of Hooks is keeping related functionality together. With React Classes the code was spread out over mount/update/unmount event callbacks on the instance, which got really messy.
|
# ? May 14, 2022 00:03 |
|
I'm coming around to the idea that maybe, just maybe, these hard to understand data structures could be designed better rather than I'm too stupid to understand them
|
# ? May 14, 2022 01:46 |
|
Is there a limit to how many joins is too many? We're doing aggregates that pull in 7 tables as part of a cte
|
# ? May 14, 2022 01:46 |
|
query optimizers are so good nowadays that there is a material chance the joins will not be bottleneck
|
# ? May 14, 2022 02:06 |
|
if you've got joins that are just looking up a single entry via an index then it'll probably be just dandy if you've got seven wild-rear end joins that have to reduce to table scans because there's no other way to solve the constraints then you're gonna have a bad time, lol
|
# ? May 14, 2022 02:54 |
|
Yeah like Jobor and bdid said, assuming you have sane indexes and don't have to do full table scans for everything, 7 joins is no problem for a modern optimizer. As an example, PostgreSQL will exhaustively find the most optimal join order via dynamic programming for queries with up to 12 joins (by default at least, uses "geqo_threshold" which you can configure - on line 3293 in the source here you can see the check where the optimizer chooses between the "standard" algorithm and geqo). Depends on what you're doing with that CTE though, if it's recursive and has enough depth then you might have a really bad time ThePeavstenator fucked around with this message at 03:24 on May 14, 2022 |
# ? May 14, 2022 03:22 |
|
if you run into problems just do some of the joins, make a temp table, then do the rest of the joins
|
# ? May 14, 2022 03:42 |
|
amazon has a postgresql saas, now google has a postgresql saas, and I just found out about a third service among probably a lot of other companies doing it too. I'm not a MySQL/MariaDB superfan but I'm curious. Is MariaDB just not catching up? Or are they attempting to corner a different market or utilize a different philosophy? Maybe I'm stuck in the last decade but I was under the impression that they were about neck and neck in popularity for enterprise deployments that favor open source sql-based RDBMS.
|
# ? May 14, 2022 04:33 |
|
postgres has always been superior in features but now it caught up in performance. meanwhile mysql got the stain of oracle
|
# ? May 14, 2022 04:36 |
|
mysql’s big advantage was that it’s a lot easier to implement a mostly working good enough rdms than a correct one, but eventually Postgres progressed from a theoretically better but impractical rdbms to an actually better one if the oracle thing hadn’t happened then perhaps mysql could have continued to outpace pgsql, but it really sucked a lot of the air out of the room
|
# ? May 14, 2022 05:53 |
|
|
# ? Jun 8, 2024 06:45 |
|
Plorkyeran posted:mysql’s big advantage was that it’s a lot easier to implement a mostly working good enough rdms than a correct one, but eventually Postgres progressed from a theoretically better but impractical rdbms to an actually better one replication is still a sore spot in postgres, but just about everything else about it absolutely destroys mysql these days. especially the stupid but amazing poo poo like indexed json fields.
|
# ? May 14, 2022 06:05 |