Fork me on GitHub
Alex Miller (Clojure team)01:04:48

well CLJ-2112 has a prototype of that :)

clojure-spin 25

Is this a reference to the Rush album? I hope so

Alex Miller (Clojure team)14:04:54

it is merely a very happy coincidence that allows me to trivially remember the ticket number

😄 3

I've learned of the awful world of Atlassian. Weve started using it for OpsGenie. Just trying to navigate their site and logins and stuff makes me want to throw the computer out the window.


When did everything become so bad? Computers nowadays feel like devices to slowly kill you via death by a thousand cuts, everything is slow, buggy or just outright broken. Nothing is documented, nothing works and it's so all bloody frustrating. When did this become acceptable! I should go live in the woods:confused:


I don't understand it. Atlassian are a hugely valuable company yet the stuff they make is complete garbage. I just can't understand how enough people used this crap and thought it was acceptable that the company ended up being successful.


I think at some point everyone else has been taken over by pod people.

Dimitar Uzunov14:04:48

because the people who buy atlassian rarely are the ones using it 😄

🎯 9
Dimitar Uzunov14:04:57

what I would love in similar software is good APIs so that I can for example do my own cli or simplified frontend without having to use their UI messes


yeah, i've been trying to interact with the opsgenie api for a few days. The documentation is absolutely terrible

Dimitar Uzunov14:04:03

I have a small POC of a dashboard for JIRA.. I gave up doing that because I ended up reimplementing their UI, it is just too inflexible.


I want to check if an alert was created, i get 3 fields back: IsSuccess: true/false Success: true/false State: "Alarm was created" (but could be other strings...) WTF! I want to know 1 piece of information, but this is conveyed through 2 bools and a string...

😅 6

what if one field is true and the other is false...

Dimitar Uzunov14:04:06

if you want to have fun try their DB tables


These systems are incredibly configurable, and they must get suggestions for enhancements from dozens of heavy users. Part of what you are seeing might be a poorly configured instance of JIRA. Part of it is probably the accretion of features without saying "no" often enough for a dozen years of the lifetime of a popular product.

👍 6
Dimitar Uzunov16:04:13

Well I'm not sure about that


When you see a snapshot of the end result of such a long process, it often looks like a mess. If you saw each individual decision as it was made, each one of them might look much more reasonable than the final system does.

Dimitar Uzunov16:04:40

Usually Atlassian say no - and your option is to buy an addon from the marketplace


It can be hard to say no to feature additions, especially from large customers paying you a lot of money.

Dimitar Uzunov16:04:06

And if the addon gets commercial success then then they will consider adding it


OK, so you probably have a lot more experience with JIRA in particular than I do. So if you buy 3 add-ons developed by different teams, there is probably absolutely no guarantee that they will be smoothly integrated with each other?

Dimitar Uzunov16:04:24

Of course, they might even conflict

Dimitar Uzunov16:04:04

I've used the server edition, don't know if cloud is better on that front

Dimitar Uzunov16:04:26 here is one example of a feature that they haven't implemented since 2016


I mean, yeah they might have lots of features but there's just no excuse for garbage like this:


In a product that we are going to be paying for


We use the cloud stuff at work and it’s slow, and clunky in terms of the UI/UX. I got selected for a random user survey via Zoom last year and some of the stuff I showed them as being broken, the Atlassian person didn’t even know about — and I will say almost everything I raised during that call has since been fixed so I’ll give them props for that.


But when a software product is popular with (enterprise) companies, it usually ends up with a giant mass of options and features and flags and configuration — and there are often combinations that don’t quite work properly — because they’re getting pulled in every direction, and those corporate purse strings are pulling strongly.


So I sympathize with their position (I worked at Macromedia/Adobe so I’ve been on the other side of the “corporate user” fence) but, yeah, it does tend to lead to bloat and a sub-par UX for “more discerning” users.


For us, the fully-integrated nature of JIRA/BitBucket, connected to Slack, and with “good enough” integration into VS Code makes it worth it overall (but it’s a bunch of trade offs).


I was wondering today if getting into game dev would be an option


It’s the same reason we’re an Office 365 shop: for the price, we get a decent bundle of integrated services that, across the whole company, serve enough people “well enough”.


They can't possibly have to deal with the mass of garbage that most devs have to deal with nowadays, poorly documented open source crap that half the times doesn't work, crap like Atlassian etc.


Not even mentioning Azure! and the Azure portal!


To be fair, at least a lot of this corporate “crap-ware” has plenty of documentation most of the time. A lot of OSS projects have much worse documentation.


I was on a call with someone from MS tech support for one of teh Azure services we use. The service just randomly keeps... stopping... Just stops... And we have to go into the portal and reset it... I was on the phone with him for a while and he forgot he was sharing his screen. He started googling my issue and he ended up on stack overflow!


I was like dude, I've tried googling.... Your supposed to be the expert!

Dimitar Uzunov16:04:57

I've had this happen to me, got redirected to stackoverflow


Most large companies’ tech support is aimed at the vast majority of their less technical users — the ones who have no chance of figuring stuff out. You’re an “above-average” user.

Dimitar Uzunov16:04:09

And someone from Microsoft answered there

Dimitar Uzunov16:04:49

The other thing that happened was I was using cosmosdb and it suddenly stopped working on new years eve

Dimitar Uzunov16:04:26

Turns out the mongo api they reimplemented was changed to a newer version

Dimitar Uzunov16:04:32

Without me doing anything

Dimitar Uzunov16:04:01

Got got in touch and they actually hooked me up with one of the engineers writing CosmosDB

Dimitar Uzunov16:04:30

And he was super kind and understanding, but his solution was to try to help me change our Backend

Dimitar Uzunov16:04:47

So.. we just switched off cosmosdb and ran Mongo in a VM

Dimitar Uzunov16:04:03

And I went to enjoy my NYE..


I had an issue with my Surface Laptop 3 that involved the CPU overheating and getting throttled after a firmware update. I spent a week going back and forth with MS tech support trying to convince them it was a firmware bug — similar to the one that had surfaced eighteen months earlier (that took them six months to fix). He wanted me to do all sorts of reinstallation and reformatting — and would not budge from his insistence that I stop using Insider builds… a few weeks later, there was a new firmware update and — guess what? — it fixed the problem!


we experimented with cosmo db a couple of years ago, we didn't think it was production ready yet


But speaking of Azure, we run RabbitMQ via CloudAMQP hosted in Azure. One day it went down for about 40 minutes, which isa LONG time for us. Here was the email exchange with CloudAMQP CloudAMQP: "Azure preformed maintenance on your instance, involving a restart" Me: "And it took 40 minutes?" CloudAMQP: "Yes, it started at 13:49:30 and was reported finished at 14:27:02" Me: "Any reason why it took 40 minutes?" CloudAMQP: "I was migrated to a new underlying host which is why it took some extra time. I can't find anything else in the logs." Me: "40 mins is quite a lot of down time." CloudAMQP: "That's why we don't recommend Azure. AWS will notify about two week ahead of scheduled maintenance, and you can manually reboot the instance before that, and GCP can even live migrate VMs between physical hosts causing 0 downtime during maintenance. Azure does neither." Great! MS just decided to upgrade something without telling us, and knocked out instance out for 40 minutes!

Dimitar Uzunov16:04:15

Well you are not entitled to an SLA unless you use an availability set with the machines in different update and fault domains

Dimitar Uzunov16:04:38

What Sean mentioned is definitely the case with Azure - they outsource most support to organizations who do not have any special training or access to help you, they have access to the same docs you can get from searching the web


AWS might be just as bad then


for support i mean

Dimitar Uzunov16:04:45

A better channel for Azure might be to call your Presales/Evangelist/etc. and they might be able to give you a pointer or reach out to the product guys

Dimitar Uzunov16:04:55

Everything on best effort of course

Dimitar Uzunov16:04:54

What you see on GitHub, Stackoverflow and their doc sites is all the documentation there is

Dimitar Uzunov17:04:57

And even thought it looks like community docs it is mostly written by MS employees.. kinda like the Chinese volunteers in Korea..


FWIW, AWS instances can — and do — randomly lock up completely and become unrecoverable and you have to build and deploy a new instance. Cloud services all tend to be like that.


Yup, we have ~200 ec2 instances and we see some kind of failure around once a week... VMs just stop responding, ephemeral drives bite the dust, network partitions, etc. It's normal and unsurprising that once you get beyond small scale you start to see some cracks.


Couple thoughts here, I've had good experience with AWS support. I've found what seems like bugs (500 exceptions), and support have been able to give technical answers about areas that aren't supported that we're hitting. We've also had performance issues with RDS dbs in prod with AWS, and support were able to diagnose the issue down to the host node the DB was on.


EC2 instances definitely fail, go offline, etc. The AWS argument is that this is perfectly normal in any cloud environment, and you should be using load balancers etc. to cope with that. I'm surprised the rate is as high as 1/200/weekly though. I've not run at that kind of scale before though. I did have a non-HA Datomic DB go down on me once though as a result of an EC2 lock up :)


I also was forced to use Atlassian products in some of my jobs. Every time, the excuse is the same: "it's just poorly configured...". So I call it a downright lie - their products are bad, and people use the "configuration card" to reason about


I mean, bitbucket didn't allow me to create a pull request if the branch changed when I opened the page, and also lost all approving status when the branch is updated; its UI is inconsistent, sometimes it does syntax highlight, others don't.


It also sometimes will allow me to start two comments, and other times it'll close the others without telling me.


Jira is worse - create a card, task, subttask, and story were ALL in different places (context menu, submenu, other screens, etc).


I… might just get one today 😅

🚀 3

There's at least one interesting review,

I would marry this if it was allowed and I wasn't already married

😅 3

They look great, just ordered one!


Just got mine. Works as advertised so far, erases pretty clean!


Found this to be an interesting option. Been using Miro lately to organize loose thoughts before putting more structured ideas into my zettelkasten, but this might be a bit more natural. We’ll see!

replied to a thread:

There's at least one interesting review,

I would marry this if it was allowed and I wasn't already married

😅 3