@sir straight up slap in the face to the bloaty forges.

i remember when we used to care about site speed and usability.

can actually believe how horrible bitbucket is, that atlassian slow as molasses mentality.

@sir curious, where was this tested? locally on a physical pc, or on a vps? if a vps, did it have any location advantage? (ie same provider/datacenter/network as sourcehut or any other forges)

@mendel on builds.sr.ht, which is in the same datacenter as git.sr.ht. However, lighthouse's simulated latency accounts for this, and the MTR reports are not especially generous to sr.ht:

builds.sr.ht/~sircmpwn/job/188

@sir it's my first time seeing mtr, but it seemed to me that sourcehut did have the best result, with only gitlab coming close. maybe it should be mentioned in the disclaimer or something.

at any rate, it is really cool that it will rebuild periodically and stay updated! it would be nice to have some graphs once enough data is collected, so we can see the trends of who focuses on what (performance vs features)

@mendel aye, but again, lighthouse accounts for the latency, so it doesn't affect the results.

@mendel oh, and I had a friend run the same test suite from europe, with better peering to some of these hosts - and the results were not markedly different.

@sir that's reassuring. maybe someday we will have distributed testing xD

@sir I noticed some misspellings on the page

s/comparible/comparable

"Bitbucket: No comparible", should be "Not" instead of "No"

@tristan957 @sir > should be "Not" instead of "No"


Really? It seems like 'No' would have made more sense in context:



* Bitbucket: Not comparable, public JIRA instances are currently known to the authors of the performance tests

v.

* Bitbucket: No comparable, public JIRA instances are currently known to the authors of the performance tests

@kick @tristan957 doesn't matter in any case - I found one to compare with and the next report will be corrected

@sir

> Software Forge Performance Index forgeperf.org

I'm not surprised at all to see SourceHut at the top – congrats.

I *am* a kind of shocked at how well GitHub is able to do, given that it's a Ruby project under heavy load. So, congrats to them too, I guess?

@codesections GitHub has the advantage of heaps and heaps of money 😄 but they have room to improve, and so does SourceHut

@sir

> GitHub has the advantage of heaps and heaps of money 😄

Yeah. I tend to be pretty skeptical of attempts to solve performance problems by throwing money (which usually means developers) at the problem – that often *increases* complexity in unhelpful ways. But I guess this goes to show that isn't always true.

> but they have room to improve, and so does SourceHut

Well maybe, but in at least 2/3rds of the categories, any improvement for SourceHut will be beyond Lighthouse's measurement!

@sir something is weird about the sort order in the “Log (worst case)” section?

@kensanata it looks right to me? It's sorted by performance score, best to worst

@sir oh, I see – the tables can be scrolled horizontally. When I looked at it, there was no indication of that.

@kensanata ah, you're on mobile :) imo this is a failure of modern mobile browsers refusing to show scroll bars on things

@sir Good old Atlassian … leaving the Doherty Threshold in the dust and proud of it!

Sign in to participate in the conversation
Mastodon

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!