I think the worst thing you could do in this situation is put the redesign people on the infrastructure team

Exactly. The kind of comment you're replying to is what I expect on a forum where gamers complain about companies not putting content artists on "fixing the servers", not Hacker News.

This. It’s not as if Slack is a tiny startup with three engineers. They’re a large company now of 2,700 or so, with nearly 1,000 working in engineering. At that scale, I am certain they can manage a UX redesign in parallel with any amount of back end engineering required.

I thought they already were, given what is happening.

Sorry, couldn't resist making the joke.

Or just stop the nonsense redesigns in general. Moving things around every few months doesn't benefit anyone.

If the new design was clearly superior, fine, but then if it was clearly superior, they wouldn't do it again a few months down the road.

Maybe then Slack could also put resources into creating native apps. Electron supposedly streamlines development and allows teams to move quickly—but I'd rather they move slowly and make performant software.

Curious what perf problems you’ve run into on the frontend (I.e. things electron would impact)? It doesn’t seem to me like they do anything complicated enough to take time.

I spend most of my time on a computer with a 4790K and 32 GB of ram. There, admittedly, I don’t have any performance issues—but that’s because I have a ridiculous amount of overhead for something like Slack!

I sometimes use lower-end machines to test stuff, like a 2011 Macbook Air with 4GB of memory. There, I have all manner of problems—the fans spin up, and the whole Slack interface becomes extremely slow to respond. Simple things like changing channels and loading threads take seconds at a time.

These computers are inherently slower by modern standards, but there’s simply no reason Slack should require so many system resources!

I understand where you’re coming from, but I also think that having a generally-reasonable-ish response time on decade old hardware that was considered lightweight-tier even at the time is not as awful as you make it seem.

I'm on a 2017 MacBook Pro right now with a 300Mb connection, and using the web interface takes multiple seconds to change channels, has a > 1 second delay in typing (resulting in missing keystrokes). My workstation isn't a particularly pleasant experience and it's _much_ beefier than this machine. My partner uses MS teams on a beater of a windows laptop with at most 4GB of ram and it performs substantially better.

I've made multiple support complaints to slack about their memory usage, there's been times where it's been using Gigabytes of memory on my workstation, and I've actually had to kill slack to compile on more than one occasion.

> My partner uses MS teams on a beater of a windows laptop with at most 4GB of ram and it performs substantially better.

So not an electron issue, an engineering issue. That's I think at the root of all these "electron is bad" comments... poorly engineered code is slow. It will be slow if it's C++ or JS or Perl. But well engineered code is fast. And it can be fast in C++ or JS or Perl.

I don't disagree, but starting in electron is going to make your life much more difficult. Well engineered electron-based code may still be slower than shoddy C++

As an example, I've been playing with web frameworks recently. Asp.net running on dotnet core is roughly 10 times faster than node running express. While well engineered js might not get much slower, you'll never close that gap.

AIM, ICQ and Yahoo Chats would work fine on that machine what features does Slack have that requires the additional resources?

I'd expect to have the same experience on most $300 Windows laptops, though. And even on better computers, what is that doing to electricity usage and multi-tasking performance?

FWIW I've had significantly better performance using the web interface, pretty much at all times. And as a bonus, you can open the same slack instances in multiple tabs and copy/paste/crosslink with relative ease.

It's still a hog of course, but it's broadly better.

I find the biggest perf issue is typing! seriously it actually lags behind my typing at an alarming delay sometimes, this is not being used on a slow or busy machine, there is something going on in slack front end that evaluates on every keystroke and sometimes it's really really fucking slow.

I completely agree. While the UI was up to each individual user when we used IRC it was extremely stable. A lot of the Slack-isms are anti patterns and I can’t help but feel schadenfreude when they publicly have problems. They’ve caused enough for me.

I'd focus on improving performance. The web version is so unusably slow that I don't know how Slack is still taken seriously as a company

How is a multi-billion dollar company incapable of implementing a web chat in a competent way?

It’s as if they’re running a gigantic-scale, totally unique collaboration platform the likes of which the world has never seen, during a period in which a global pandemic has increased the use of the platform massively, leading to scaling challenges that anyone outside of Google, Microsoft, Facebook, Dropbox, Slack and a mere handful of other giants would scarcely comprehend...

I'd give them that credit had Slack not always been a slothful sludge of molasses. For years.

Any app in which I can type faster than the characters appear has deeper issues than scaling.

> It’s as if they’re running a gigantic-scale, totally unique collaboration platform

No it’s a centralized SaaS IM/chat platform with a very slow web client and no native clients, because it’s not a standardised internet protocol surrounded by a healthy ecosystem.

And yet, a company that is about a quarter its size and entered the race 2 years later is able to beat them in features, performance and stability, even in these trying times.

All of that and they don't have Enterprise contracts with SLAs!

(I'm talking about Discord)

Anecdotally, Discord is down and unresponsive far more than Slack ever has been. I would never run Discord as the main method of communication at a large company. It might be fine for when I game, but it's unreliable for anything serious.

Here's some data to back it up:

Slack's uptime is 99.95% from August to October[1]

Discord's uptime is 99.85% from August to October[2]

[1] https://status.slack.com/calendar [2] https://discordstatus.com/uptime?page=1

Perhaps there is some regionality to the issues? I live in Brazil, and Slack is often somehow unavailable: images not loading, message delays, failure to load threads, and the built-in bot commands (not third parties) not responding. And mainly, voice calls having all sorts of connection issues imaginable.

In the last 3 months our team has had to (against company policy) fall back to Discord to be able to do our daily meetings about 6 or 7 times.

I wouldn't run Discord on a large company either, but not for stability's sake, but because their (free) screen sharing is possibly the worst if you want to share a desktop rather than a game (it is very much optimized for low delay, and the downscaling makes text unreadable).

I'm forced to use Slack for work and I notice that there's a multi-second delay between my keyboard input and when the characters show up on the screen.

I believe you, but I think you're hitting something completely unique to your setup and external to Slack. I've never seen antyhing like that, what are you running it on? Do you have ways to inspect to find where the bottleneck is?

> I believe you, but I think you're hitting something completely unique to your setup and external to Slack.

That's entirely possible, but this problem has persisted throughout multiple versions of Firefox and Fedora. For all I know, it could be due to the VPN connection, but even then, the VPN connection isn't that slow. Zoom video conferences still work reasonably well over it.

You must use a different Slack than I do, because I use nothing but the browser client and have had zero issues like this or the parent comment. Get up in arms about the outages, because that’s a bit more justified, but this feels a bit contrived?