|
DONT THREAD ON ME posted:your advice is really good and it's benefited me in the past. as you say, though, I think i'm mostly traumatized by my previous job where we didn't really have any users and we were building a very complicated product under the steering of people who were just guessing at what the users wanted. Two of my biggest projects were canned. Both were around a year long, and one was cancelled a month from shipping, the other one we pulled the plug a week from deploying (I fortunately at least had one or two other big projects that shipped properly over the last few years). The way I spin these positively is that we learned and built everything we had to, but also had the good judgement to know when to get rid of a project that would not be successful ("kill your darlings") and move on successfully. It's always a hard decision to make, and being able to agree with it and even make the call yourself, and then live with it, is what I'd say is a move that required a lot of maturity from me over time. The big regret is of course to not know how it would have possibly worked or what kind of customer feedback it could have received, but acknowledging failure and bad approaches without shoving them in production so they instantly become legacy is a good engineering decision. It's of course different if things fail because the implementation was not up to par than if the requirements were unclear or shifting too often, but experience is experience. You may want to mark failed projects as 'prototypes' in a resume if you must. You just have to have figured out some good lessons in that phase even if it never shipped.
|
#
?
Mar 13, 2019 20:21
|
|
|
|
| # ? Apr 27, 2024 15:36 |
|
|
did a small presentation on my ocr/pdf app and they were super impressed even though it crashed a bunch of times lol python 
|
|
#
?
Mar 13, 2019 20:58
|
|
|
nice, also as par of the course. when i demoed my bi dashboard (r shiny+plotly) some time ago the demo began with hard etl crash, but i had enough common sense to hardcode paths to csv backups for that scenario lol
|
|
#
?
Mar 13, 2019 21:01
|
|
|
yeah i made sure i had all the intermediate files that are generated so i could show and explain how A goes to B to C (or is supposed to -- it was the B part that was crashing because i stupidly hadnt tested it properly in a while), then skip to the next part. then D crashed but for that i could revert to a commit from a couple days ago and show that 
|
|
#
?
Mar 13, 2019 21:07
|
|
|
also it was right after i had smugly shat on "academic code" and how fragile it is omg lmao
|
|
#
?
Mar 13, 2019 21:08
|
|
|
|
|
#
?
Mar 13, 2019 21:20
|
|
|
CRIP EATIN BREAD posted:I'm not sure if it's a function of kops or kubernetes itself, but all the nodes get labelled with "failure-domain.beta.kubernetes.io/zone=AWS_AZ" that you can easily use for node affinity to a specific AZ. this is useful if you actually have persistent volume claims. that's shortcoming of EBS though, since a volume is created in an AZ and not available in other AZs. This is a newer feature of kubernetes itself. We get those labels on GCP and Azure as well. quote:what state do you need to maintain in a file? seems like you would have the same issue just using AWS autoscaling groups. this really isnt a kubernetes specific issue. It's definitely not a k8s specific issue, but I think it shows up more frequently than for bog standard ASG setups. We run redis internally and it persists the state to disk. We use a volume for this so if the pod gets moved the data isn't lost. For the most part we stay away from volumes for persistent data, but our on-premise package uses them for elasticsearch as well. sometimes you just need a persistent disk, ya know?
|
|
#
?
Mar 13, 2019 21:22
|
|
|
oh cool a giant pile of powershell scripts, xml files, access databases, excel sheets and bat files on a network share, looks like somebody has been busy making GBS threads out macros again and we'll be learning the lessons of the great RAD clusterfuck all over again in tyool 2019! ~7 years ago it would have been me doing it but now I know better and it makes me angry
|
|
#
?
Mar 13, 2019 21:23
|
|
|
DONT THREAD ON ME posted:i am going through a second level of imposter syndrome wherein i doubt my ability to ever give a gently caress about solving a business problem beyond finding whatever cute and satisfying technical solutions that i can. i am seriously considering going back to someplace like redhat where i can just write easy code at work and spend my free time programming stuff i enjoy. what i like with my job is that game designers are paid to figure all that stuff out and at worse I just have to point at corner cases they forgot to specify "what if the player does this then that and becomes unable to progress because ___" and just wait for them to figure out a solution as gameplay programmers we basically end up with a spreadsheet with things like "pressing X should make the player fart" and "additional fart sounds should be purchasable with hard currency at the ingame shop" and we just implement all that (we just have a technical validation phase first to point out all the unfeasible, contradictory or way too expensive to build stuff so that the designers can fix it first) Zlodo fucked around with this message at 21:37 on Mar 13, 2019 |
|
#
?
Mar 13, 2019 21:35
|
|
|
tfw you ask someone to submit a pr and they email you a zip file of their changes instead
|
|
#
?
Mar 13, 2019 21:45
|
|
|
Soricidus posted:tfw you ask someone to submit a pr and they email you a zip file of their changes instead 
|
|
#
?
Mar 13, 2019 21:53
|
|
|
Soricidus posted:tfw you ask someone to submit a pr and they email you a zip file of their changes instead If my team starts interviewing again, poo poo like this is why I'm going to include a section on Git. No turbo-nerd command line trivia, you can use Google and whatever tools you want, just "add a line of code, commit to a new branch, create a pull request".
|
|
#
?
Mar 13, 2019 22:33
|
|
|
ahhhhh poo poo cjs: CEO wants to add some backend dudes from India, now. I pointed out how well that worked with the Russians (not well) but he's so "stoked" about how much he is saving on these guys ("they cost like 1/5 as much as our frontend contractor"). I did manage to talk him down from letting him directly work on code for our core platform. And oh yeah this guy would report directly to CEO and not be included in any of our sprint planning. I asked that all his work be added to jira because I wasn't going to approve pull requests unless we had documentation of what the work was supposed to be, and I got a bunch of passive aggressive jira tickets with 1-line descriptions like "build page per slack discussion" I think I got like 3-4 months before this startup runs out of money... we'll see.
|
|
#
?
Mar 13, 2019 22:41
|
|
|
|
|
#
?
Mar 13, 2019 22:44
|
|
|
DONT THREAD ON ME posted:your advice is really good and it's benefited me in the past. as you say, though, I think i'm mostly traumatized by my previous job where we didn't really have any users and we were building a very complicated product under the steering of people who were just guessing at what the users wanted. c tp s: 4 rejections so far, only 1 process outstanding until I gotta go starting applying from the beginning again
|
|
#
?
Mar 13, 2019 22:59
|
|
|
Finster Dexter posted:ahhhhh poo poo start your job search now, if you haven't already
|
|
#
?
Mar 13, 2019 23:16
|
|
|
Finster Dexter posted:ahhhhh poo poo you're already dead 
|
|
#
?
Mar 14, 2019 00:29
|
|
|
it was "suggested" the other week that maybe, just maybe paying absolute bottom dollar for offshore devs might not be working out so well because best case if they turn out to be any good they spend 6-12 months getting trained then gently caress off and we start all over again plus their management is so incredibly hierarchy driven that anyone who shows any talent and doesn't bail is a threat to their boss and either gets shunted out or is ruthlessly undermined. a lot of the management tier also have zero technical knowledge and spend all their time trying to do as little as possible so there's no chance that anything could happen that would make them look bad so anyway as i said Powerful Two-Hander posted:you're already dead
|
|
#
?
Mar 14, 2019 00:34
|
|
|
necrotic posted:This is a newer feature of kubernetes itself. We get those labels on GCP and Azure as well. there's an EFS volume provisioner kicking around in k8s-incubation if volume content living on the other side of an NFS target is ok. I'm intending to tinker with it at some point and see how it fares but I haven't had time.
|
|
#
?
Mar 14, 2019 03:20
|
|
|
Finster Dexter posted:ahhhhh poo poo lol you poor bastard. also man jira, i loving hate that thing. everyone tries to use it but its so godawful slow that if i dragged all my dumb cards around it would be all im doing. people are starting to learn, the other day I overheard "don't bother assigning <CRIP EATIN BREAD'S FIRST NAME> any tickets, he's never going to loving use it" so my efforts to actively dodge jira aren't going unnoticed.
|
|
#
?
Mar 14, 2019 04:55
|
|
|
man, I'd be loving lost if I didn't use jira
|
|
#
?
Mar 14, 2019 05:01
|
|
|
Soricidus posted:tfw you ask someone to submit a pr and they email you a zip file of their changes instead same but people who are ostensibly sysadmins who send screenshots of text-only terminal contents. sure do love manually transcribing poo poo elsewhere and only having as many logs as fit in whatever window size goofus prefers for their terms
|
|
#
?
Mar 14, 2019 05:54
|
|
|
akadajet posted:man, I'd be loving lost if I didn't use jira protip: just put a todo.txt in the repo ya dingus
|
|
#
?
Mar 14, 2019 08:36
|
|
|
also im trying to get readthedocs to build my docs but it fails on tesserocr because the non-python libraries that it depends on are missingcode:e: aha, gonna try this: https://docs.readthedocs.io/en/stable/faq.html#i-get-import-errors-on-libraries-that-depend-on-c-modules e2: 
Carthag Tuek fucked around with this message at 08:59 on Mar 14, 2019 |
|
#
?
Mar 14, 2019 08:40
|
|
|
Krankenstyle posted:protip: just put a todo.txt in the repo ya dingus oh sweet summer child, work in corporate and you won't have time to make a todo. I'll call a meeting in two weeks with it architecture, infrastructure and a project manager, then we can discuss todo file structure and layout.
|
|
#
?
Mar 14, 2019 09:23
|
|
|
why is jira so slow anyway, we keep getting told we have to migrate instances because "the current one is overloaded" but that implies it's due to the number of actual jiras which is ridiculous unless the underlying data storage/model is complete garbage
|
|
#
?
Mar 14, 2019 09:24
|
|
|
Krankenstyle posted:also im trying to get readthedocs to build my docs but it fails on tesserocr because the non-python libraries that it depends on are missing I had a very similar issue when working with python in azure web apps. When installing python libraries with c components that need compilation the compiler would refuse (because its a c compiler in windows), but only when and if the installation was prompted by a CI tool. manual package installation worked fine.
|
|
#
?
Mar 14, 2019 09:26
|
|
|
Boiled Water posted:I had a very similar issue when working with python in azure web apps. When installing python libraries with c components that need compilation the compiler would refuse (because its a c compiler in windows), but only when and if the installation was prompted by a CI tool. oh god yeah ive been there. its such finicky bullshit and super annoying to debug  esp when developing on linux/mac and deploying on windows Carthag Tuek fucked around with this message at 09:54 on Mar 14, 2019 |
|
#
?
Mar 14, 2019 09:50
|
|
|
This is literally what docker is for
|
|
#
?
Mar 14, 2019 10:21
|
|
|
gonadic io posted:This is literally what docker is for i aint no ops guy, I'm too old for that poo poo
|
|
#
?
Mar 14, 2019 10:51
|
|
|
Krankenstyle posted:i aint no ops guy, I'm too old for that poo poo treat docker like a make file that produces a static binary image you can run in the cloud. so if you can build it on your dev machine you can run it where you need
|
|
#
?
Mar 14, 2019 11:08
|
|
|
like java promised I guess
|
|
#
?
Mar 14, 2019 11:09
|
|
|
put java in docker for double portability
|
|
#
?
Mar 14, 2019 11:16
|
|
|
Powerful Two-Hander posted:why is jira so slow anyway, because Atlassian makes an absolutely colossal amount of money so the incentive to "make it faster" does not exist. friend of a friend is the PM for confluence - they apparently rewrite the front end very often b/c they have nothing better to do. the answer to JIRA woes is: use a hosted provider or hire a team of 3-4 people who do nothing but janitor jira. And buy really, really fast disk.
|
|
#
?
Mar 14, 2019 11:21
|
|
|
my stepdads beer posted:like java promised I guess i wonder how slow a docker container on an ancient nokia flip phone would be. probably still faster than javame
|
|
#
?
Mar 14, 2019 11:22
|
|
|
noone in my job wants to manage swarm or k8s and it's becoming a real fissure between the people who want to run docker but don't want to operate that poo poo and the people that charge ahead anyway, leading to a bunch of half-built clusters everywhere
|
|
#
?
Mar 14, 2019 11:31
|
|
|
i added a Set<Butt> to a Turd class and suddenly hibernate tries to read the whole database when i query for a single Turd  thankfully it only broke staging and not production and i was able to kludge it out because the Set<Butt> was only ever read and never written to and also it was only used in a feature that was broken anyway
|
|
#
?
Mar 14, 2019 11:41
|
|
|
Feisty-Cadaver posted:the answer to JIRA woes is: use a hosted provider or hire a team of 3-4 people who do nothing but janitor jira. And buy really, really fast disk. yeah we have good jira janitors and our jira works decently fast, dunno what hardware it's running on tho
|
|
#
?
Mar 14, 2019 11:57
|
|
|
my stepdads beer posted:treat docker like a make file that produces a static binary image you can run in the cloud. so if you can build it on your dev machine you can run it where you need hrmmm i might look into it i guess Wheany posted:i added a Set<Butt> to a Turd class and suddenly hibernate tries to read the whole database when i query for a single Turd well clearly a turd cant have many butts!!! even if its some kind of ))<>(( situation, the turd is only in one butt at a time!
|
|
#
?
Mar 14, 2019 13:52
|
|
|
|
| # ? Apr 27, 2024 15:36 |
|
|
Krankenstyle posted:hrmmm i might look into it i guess for the simplest use case, some crud app you need a Dockerfile in the source dir containing: FROM microsoft/python-bullshit # get images from dockerhub COPY . . # just move your entire source directory into the image RUN apt-get upgrade && apt-get install -y my butt foo # install your poo poo that isn't in the base dockerfile above, avoid multiple RUN if you can and just use a bunch of && EXPOSE 8080 # or whatever you need people to call you on. skip if you're not running a web server CMD ["python", "myapp.py"] # whatever you use to run it normally on the server, the dumbest way to run it is to checkout the repo, then: docker build . -t "my-butt:1.0" docker run my-butt:1.0 \ -p 8080:8080 # map container port to host port \ -e "MY_ENV_VAR=secret-pass" # you can use env files etc too \ -d # run in background but ideally you would build in ci and then have some container registry (we use GCP's) to hold the image until the prod server pulls it. Try to avoid putting secrets into the container itself and instead pass using docker run. gonadic io fucked around with this message at 14:14 on Mar 14, 2019 |
|
#
?
Mar 14, 2019 14:04
|
|