|
Share Bear posted:it's the 2nd, and it sounds like none of you have actually done customer facing support, people never read errors to begin with, cmon nah, when i did customer support i was plenty happy to explain basic poo poo or stuff that was obvious from the error. usually easy tickets, and solving tickets was my primary job. it's not the customers' primary job to understand our software, and plenty of fancy IT tools are very user-unfriendly. they're paying for something that has support so they don't have to think as much about it when members of our own org can't learn how navigate issues or questions they don't have an immediate solution to re our own software or the domain it exists in, then i get spicy. you've got "senior" in your title and you're paid to understand this stuff so our customers don't have to. learn to read an RFC, review our code, or use diagnostic tools to get a baseline understanding of something, don't just immediately waltz into the biggest slack channel you can and @here that something doesn't work
|
#
?
Nov 5, 2021 20:17
|
|
|
|
| # ? Dec 3, 2024 15:28 |
|
|
12 rats tied together posted:use supervisord for that imo. it has a web interface and an xml rpc thing too thanks for the recommendation. i'd like to try that out, but we have other process deploy/monitoring stuff on the distributed monolith process i busted that won't take kindly to it, had to use shell. maybe if/when i get to build something new rather than maintain something old. quote:when members of our own org can't learn how navigate issues or questions they don't have an immediate solution to re our own software or the domain it exists in, then i get spicy. you've got "senior" in your title and you're paid to understand this stuff so our customers don't have to. learn to read an RFC, review our code, or use diagnostic tools to get a baseline understanding of something, don't just immediately waltz into the biggest slack channel you can and @here that something doesn't work While I do have somewhat similar expectations of coworkers, I find it more productive to lead with empathy & guidance in the long run. Sometimes people don't have context for what or where to even begin, it happens to everyone. I am also a huge idiot so I think going with "treat others how you would want to be treated" is ok even with senior people. me posted:when it becomes a habit, when a person cannot work without bugging you, then there's an issue the line for habit is arbitrary but you can usually tell when its a problem asking Qs ain't a crime Share Bear fucked around with this message at 21:15 on Nov 5, 2021 |
|
#
?
Nov 5, 2021 21:12
|
|
|
CMYK BLYAT! posted:when members of our own org can't learn how navigate issues or questions they don't have an immediate solution to re our own software or the domain it exists in, then i get spicy. you've got "senior" in your title and you're paid to understand this stuff so our customers don't have to. learn to read an RFC, review our code, or use diagnostic tools to get a baseline understanding of something, don't just immediately waltz into the biggest slack channel you can and @here that something doesn't work but did it work
|
|
#
?
Nov 5, 2021 21:19
|
|
|
don’t dox me
|
|
#
?
Nov 5, 2021 21:41
|
|
|
hobbesmaster posted:but did it work of course not, it's software you have to specify what's not working tho, even though "all of it" isn't entirely incorrect
|
|
#
?
Nov 5, 2021 22:58
|
|
|
Share Bear posted:maybe if/when i get to build something new rather than maintain something old. 
|
|
#
?
Nov 6, 2021 00:36
|
|
|
in which companies discover the limits of agility Complexity is killing software developers “The process of application development is simply too fragmented at this point; the days of every enterprise architecture being three-tier, every database being relational, and every business application being written in Java and deployed to an application server are over,” wrote RedMonk analyst Stephen O’Grady in a 2020 blog post. “The single most defining characteristic of today’s infrastructure is that there is no single defining characteristic. It’s diverse to a fault.”… Or, as the ex-Tumblr CTO Marco Arment wrote back in 2015, “Web development has never been more complicated or convoluted than it is today due to the sheer quantity of tools (and their rapid rate of change) involved in most modern web-dev environments.” One of the consequences of the tried-and-tested approach cloud vendors have taken to product development—small, independent, two-pizza teams building services in response to customer demands—is that developers have been largely empowered to choose how to assemble this multitude of building blocks together in a way that delivers business value. “You are like a kid in the candy shop in the cloud,” said Camille Fournier, head of platform engineering at financial services firm Two Sigma, in an interview with InfoWorld. “But as you grow and try to make things fit together, the complexity absolutely multiplies.” Which leads many to question whether this level of choice is a net positive for the average software developer. Or, as O’Grady concluded in that 2020 blog post, “The complexity inherent to a huge catalog of available services can become, in certain settings, less a strength than a liability.” This growing level of complexity has led many organizations to adopt a central platform model, where an internal platform team is tasked with vetting the tools most required by engineers, building templates, and plotting golden paths to ease their journey into production, while also centralizing functions like financial operations, security, and governance to ease the cognitive load on individual developers.… “The idea behind having golden paths is not to limit or stifle engineers, or set standards for the sake of it. With golden paths in place, teams don’t have to reinvent the wheel, have fewer decisions to make, and can use their productivity and creativity for higher objectives. They can get back to moving fast,” Spotify product manager Niemen wrote. The problem is, “developers love reinventing the wheel. Nothing satisfies me more than building a better mousetrap,” consultant Simpson said. But in a world where a lot of the answers are sitting right there on Stack Overflow, is this the best use of developers’ time? tl;dr after years of leaving developers to fill in the platform gaps, companies realize that they've diverted resources from the core product
|
|
#
?
Nov 6, 2021 04:54
|
|
|
far be it from me to point out that this is the culmination of the Microsoft antitrust and open-source dreams of the late 90s
|
|
#
?
Nov 6, 2021 05:03
|
|
|
embrace, extend, extend, extend, extend, .net, extend, extend, .net core, extend, vs code, extend, extinguish self
|
|
#
?
Nov 6, 2021 05:34
|
|
|
Gazpacho posted:far be it from me to point out that this is the culmination of the Microsoft antitrust and open-source dreams of the late 90s nothing defines anti-microsoft more than javascript and the web. all the disasters in modern development can be traced directly back to open sores/anti-microsoft free (as in no value) software
|
|
#
?
Nov 6, 2021 05:37
|
|
|
so not a nadella fan then shags?
|
|
#
?
Nov 6, 2021 05:40
|
|
|
my money for the terrible programming paradigm of the late 20s is going on going back to bafflingly constructed dodgy lightweight unix-philosophy-esque applications in C, but this time written using github autopilot and future workalikes (because you know they're coming)
|
|
#
?
Nov 6, 2021 05:44
|
|
|
CRIP EATIN BREAD posted:I see this a lot and it's really starting to drive me nuts. Last night I helped a client with some SSL cert poo poo they couldn't figure out , just because I felt bad that if it broke a lot of people would be going through some health care nightmares because a cert expires in a few hours. There’s some quote about average one bug per line of code. I have at least two. On the SSL front I have a 10 year old system with regularly updated certificates for a different domain on a system that only needs to be running HTTP. There is a F5 in front that adds SSL and connects to port 8085 to avoid “standard ports”. DBs are currently an ongoing issue, I just don’t know so how many things can be wrong. Biggest wonder is a screen of white space stripped SQL that creates views for a simple two table relation. More atypical is 40+ auto-commit SQL queries to update a single data entry. And that being called every few minutes by 100+ clients. MrMoo fucked around with this message at 15:27 on Nov 6, 2021 |
|
#
?
Nov 6, 2021 15:24
|
|
|
hi i work on a team that has literally dozens of codebases deployed to tens of thousands of servers with only a minimal set of common characteristics so every time something doesn't work outside of the half-dozen services i'm hands-on with i have to go ask somebody else to help me 
|
|
#
?
Nov 6, 2021 16:23
|
|
|
monolith as a design pattern is fine, its not a point on an axis of "monolith to distributed", which also isnt a proxy for "bad to good". its not even true that you must split a monolith apart at a certain level of scale the technology is fine, the problem has always been people unhinged enough to accept positions as "directors" of anything are rarely competent enough to direct
|
|
#
?
Nov 6, 2021 17:22
|
|
|
bob dobbs is dead posted:so not a nadella fan then shags? not a fan of all the electron based garbage microsoft is putting out
|
|
#
?
Nov 6, 2021 18:11
|
|
|
12 rats tied together posted:monolith as a design pattern is fine, its not a point on an axis of "monolith to distributed", which also isnt a proxy for "bad to good". its not even true that you must split a monolith apart at a certain level of scale the other factor at play is persistence vs application you'll 100% outgrow a single db (in terms of vertical scaling) before you outgrow the ability to throw compute at a monolith if you've architected it with any semblance of competency thinking about monolith vs microservice is more of an exercise in domain thinking and separation of data so that you don't slam a db because you've written a few bad queries on an extremely hot path microservices enforce that isolation and punt the blocking/slowdown/whatever you have in place to prevent overloading a db to a client and server that can understand that process rather than just yoloing queries at a db where the only way for a client to know if the db is being crushed is if queries take too long they're not essential in the slightest if you've designed your application to use multiple persistence layers for separate concerns and have your data modeled in a sane way
|
|
#
?
Nov 6, 2021 18:14
|
|
|
Blinkz0rz posted:
Don't both stackoverflow and let's encrypt run on (approximately) one main db + some read replicas and specialised caches for search and stuff? I agree with your point though, since running 100 instances of your monolith doesn't seem that different to running 10 instances of 10 different microservices (actually you're probably running like 50 of each because they all call each other)
|
|
#
?
Nov 6, 2021 18:53
|
|
|
Shaggar posted:not a fan of all the electron based garbage microsoft is putting out vscode is passable, and hopefully microsoft notion clone will be that as well
|
|
#
?
Nov 6, 2021 18:53
|
|
|
cinci zoo sniper posted:vscode is passable, and hopefully microsoft notion clone will be that as well you do realize who you are arguing with and what his gimmick is, right?
|
|
#
?
Nov 6, 2021 18:55
|
|
|
vscode is bad, but its still the best text editor available on linuxes
|
|
#
?
Nov 6, 2021 18:59
|
|
|
the best graphical text editor is gedit with the terminal and draw spaces plugin. Vim or nano for the cli. Jetbrains products if you need an IDE. 
|
|
#
?
Nov 6, 2021 19:01
|
|
|
its fun when people get shaggard imo. reminds me of people getting taint reaped in ages past.
|
|
#
?
Nov 6, 2021 19:01
|
|
|
(Hint: Whena post has it shouldn’t be taken seriously.)
|
|
#
?
Nov 6, 2021 19:03
|
|
|
Blinkz0rz posted:they're not essential in the slightest if you've designed your application to use multiple persistence layers for separate concerns and have your data modeled in a sane way agreed, broadly. the bad part of the monolith design is the, as a much smarter person than me puts it, the "monolith of the row". one way to approach this is to (generalizing) have fewer rows, and have those rows be narrower. another way is to consider a switch to services architecture. important to note that this is a switch, a redesign, its not an evolution or an iteration. jumping from a monolith straight into microservices, a specialized form of service architecture, is "design smell", blog ops, resume driven development, things of this nature. its not what the OP was about originally but its an interesting topic anyway my genuine $0.02 at the OP topic is that its not the role of "the supermarket of available tools and types of technology" to design a not-bullshit app for you. im sorry the guy who posted the article doesnt understand javascript and thinks front end is too hard these days but that doesnt mean javascript should never have existed (there are many other, better, reasons for that)
|
|
#
?
Nov 6, 2021 19:13
|
|
|
pointsofdata posted:Don't both stackoverflow and let's encrypt run on (approximately) one main db + some read replicas and specialised caches for search and stuff? yeah my point was that a common data store is much more likely to be a bottleneck in scaling your system than the amount of compute you can throw at a monolith. microservices are for domain separation so you can separate persistence into single responsibilities. it's one of those things that seems like overkill until you get there and the task of untangling data domains is bogged down by every ounce of your tech debt. why yes my team is going through this now, ama
|
|
#
?
Nov 6, 2021 19:18
|
|
|
The_Franz posted:you do realize who you are arguing with and what his gimmick is, right? i know, im just using it as a reason to bring loop up
|
|
#
?
Nov 6, 2021 19:30
|
|
|
I think monolith vs Microservice boils down to "how good are your Devs at abstraction". I hear some Devs here they're like "oh remember the good old days when we wanted a new feature we would just IMPLEMENT IT UGH SO FAST SO AGILE." and then they complain that they can't do it now because the codebase is restricted because people actually want to ensure that the impact of the code you're adding on a Friday night in the backend doesn't suddenly cause the elastic search logger to fail to ingest and lose all application monitoring. "Why is everything so interconnected and so poo poo" they lament as they hack together some lovely solution on our "website" that has three (3) web frameworks + legacy home brewed "framework" because over the years people were like "oh I want to add feature X imma add it in NEW SHINEY poo poo" framework then they work on maintaining some old feature and instead of trying to keep things separated they're like "imma boot NEW FRAMEWORK in OLD FRAMEWORK" which inadvertently leads to people doing "IMMA BOOT OLD FRAMEWORK IN NEW FRAMEWORK". like, dude, here you are complaining why you can't "move fast" anymore because everything is interconnected and fragile and it's impossible to upskill people into the software because good luck watching them do the software equivalent of "to understand how to bake an apple pie you first have to understand how the universe was created and the history before the first apple tree". Do you not perhaps think that "moving fast" previously might be the reason why you can't move slow anymore? Maybe if you took the time to design your poo poo so that there's some semblance of abstraction where you could say "here's component X. You use interface Y. Do not touch anything else" people could actually focus on solving their problems instead of solving your problems. Now the movement is to go to microservices because "this will reduce complexity" and we could on board people onto a Microservice faster that trying to throw them into the legacy but GUESS WHAT your Devs still suck at abstraction anyway. lovely rear end microservices all sharing the same database woo hoo. When we try to spin out a new service from the old stack due to the lack of abstraction from the old stack we can't effectively "cut" out a component cleanly. It's so bad that the lack of abstraction and interconnectedness is now bleeding into our business processes themselves making everything highly coupled. Our customer supports are like "lmao why doesn't this work" and we're like "well food doesn't work because of the history of Baz" and they'll be all like "what the gently caress does foo have to do with Baz FFS do you want me to tell the customer that their wheel sizes are determined by our submarine propeller diameters" and all we can do is shrug. What I mean to say is if your peeps are good and can design abstractions at logical parts in your software, doing a monolith is good. Doing a Microservice is also good. You should be able to easily move between both. If your peeps can't even see how their "hurp derp imma agile this poo poo by just changing EXTACTLY what we need for this specific moment only BUT I'm going to throw in whatever new poo poo I read on hacker News because I hacker" approach is strangling their future options then everything is poo poo give up and face death.
|
|
#
?
Nov 7, 2021 06:47
|
|
|
dont make me tap the old thread title
|
|
#
?
Nov 7, 2021 06:55
|
|
|
I ain't reading all that. I'm happy for u tho. Or sorry that happened.
|
|
#
?
Nov 7, 2021 10:56
|
|
|
Wheany posted:I ain't reading all that. I'm happy for u tho. Or sorry that happened. don't copy my merge approvals without my permission tyvm
|
|
#
?
Nov 7, 2021 12:35
|
|
|
DoomTrainPhD posted:the best graphical text editor is gedit with the terminal and draw spaces plugin. Vim or nano for the cli. Jetbrains products if you need an IDE. This, but Geany instead of gedit
|
|
#
?
Nov 7, 2021 16:45
|
|
|
Blinkz0rz posted:yeah my point was that a common data store is much more likely to be a bottleneck in scaling your system than the amount of compute you can throw at a monolith. microservices are for domain separation so you can separate persistence into single responsibilities. it's one of those things that seems like overkill until you get there and the task of untangling data domains is bogged down by every ounce of your tech debt. the problem is most people at the director level and up don't know / don't care about this distinction and people at the architectural level can get away with a "gently caress it" mentality for way longer than it takes for their decisions to start catching up but yeah the database is usually the problem. extreme care needs to be put into designing it and the recent explosion into cloud services has only made it that much easier to yolo your development
|
|
#
?
Nov 8, 2021 00:48
|
|
|
its a rookie mistake to think any approach to architecture about computers. microservices let you have many teams with clearly defined areas of expertise and demarkation with limited interaction. if that's a benefit then it doesn't really matter if it causes the moon to fart pancakes or gives engineers cancer, it will make a manager's time easier and is therefore inevitable. monoliths let you have a single team that is totally resistant to fragmentation or poaching, so if that is desirable, so shall it be. managers who aren't sociopaths will pick whatever was boosted in Wired or Medium last. architects don't make palaces for computers, they make recliners for suits
|
|
#
?
Nov 8, 2021 01:15
|
|
|
A dentist friend of mine who self taught python and has made some pytorch models following tutorials wants to make a career change into AI as applied to dentistry. I feel like I hear a ton about medical AI but not sure if there are any well funded startups or major companies working on dentistry. Anyone seen job reqs for something like this? Anyone know of the salaries for M.D. or D.D.S learns to code?
|
|
#
?
Nov 8, 2021 16:11
|
|
|
CarForumPoster posted:A dentist friend of mine who self taught python and has made some pytorch models following tutorials wants to make a career change into AI as applied to dentistry. I feel like I hear a ton about medical AI but not sure if there are any well funded startups or major companies working on dentistry. job requirements for being in medical tech sales is being young and attractive, i think. for actual applied ml the medical degree is a liability, unless the plan is to be the resident dentistry researcher - it definitely won’t wave away their lack of education or experience at any company with a sliver of understanding of their business
|
|
#
?
Nov 8, 2021 16:17
|
|
|
in my experience the resident researcher is also one of the people who founded the dang company so its not actually a feasible job to just go out and apply for they'd probably have better luck starting a company to solve a problem they discovered in dentistry (but--to be clear--i am not saying their luck would be good, just better)
|
|
#
?
Nov 8, 2021 18:33
|
|
|
its a good idea in principle. the tech in medicine, on the ground, is laughably terrible because there is zero empathy between the developers/data scientists and actual users, and there is also a massive funding deficit. i know someone who did a software trial in a major hospital, and it was like, "yeah this would be great but right now we can't even buy phones for every room let alone computers". dentists have a lot more money so they buy cool tech all the time but marketing is extremely fragmented like most verticals, but there is still a significant usability gap. i usually have to CJ something in my dentist's fat stack of bonzi buddies when I have a dentist's appointment. THAT SAID most start ups hate it when qualified people who know what they're talking about say "you're doing it wrong" because they can touch computers and therefore are omniscient and empathetic beyond all reason. big pharma though probably has a lot of tasty figgies -- he should ask his colgate rep or whatever the next time they're doing a junket to see if he can get an in
|
|
#
?
Nov 9, 2021 00:43
|
|
|
that there is not ai, its b2b enterprise the startups structurally cannot crack it in medicine without alternative funding arrangements because the hospital billing and renewal cycles are like every 2 years and startups cannot go that long, structurally, for doing anything. that, insane cronyism (which startups sometimes can handle but it doesnt help) and all the procurement being boomers is why emrs suck and all the rest of health care poo poo. emr bills are extortionate but cheap for b2b enterprise, the software costs are why hospitals cant afford things (and the remarkably common administrative de facto embezzling) but that comparative lack of money and pickiness also has knockon effects on only inferior touchers staying in medtech, inferior tech like mumps predominating, general kakistocracy in product quality, etc. bob dobbs is dead fucked around with this message at 01:06 on Nov 9, 2021 |
|
#
?
Nov 9, 2021 00:51
|
|
|
|
| # ? Dec 3, 2024 15:28 |
|
|
you will literally see changing 3 slocs of css taking 6 months to deploy, in a hospital environment. good luck doing high quality dev and keeping high quality touchers in that environment you will also see significant changes taking multiple years bob dobbs is dead fucked around with this message at 01:01 on Nov 9, 2021 |
|
#
?
Nov 9, 2021 00:57
|
|