We’ve been working extensively with the Legato and mangOH project. We’ve had our share of ups and downs, but to date we’ve always ended up on top. We’re so thankful for all the hard-work put into these projects. We rely on these contributions heavily to build amazing products.
Something caught my eye recently, and it smells a little bit fishy:
I was a little taken back when I saw this. Legato 18.10.2 is a full proper release, so this alpha quality software must be opt in right? No, it’s not. OK, they must have left the old Data Connection Service in place? No, it appears to be gone. OK, well maybe it will still work? No, it segfaults constantly and creates a brutal bootloop. I really hope I’m wrong about this, but based on my analysis this appears to be the case.
It’s fine that Legato 18.10.2 and the new Data Connection Service are unstable. Most new software is born this way, but I fail to understand why this saw a release. I understand that everyone inadvertently creates bugs, but it seems to me the team knew this was unstable.
If the goal of Legato is for people to adopt it, people must be able to rely on it. This created a pretty huge lapse in my trust for the project.
Fortunately, this is easily remedied. I would suggest fully adopting semantic versioning. It presents a really direct solution to this problem. For example, Legato 18.10.2 should be released as
5.10.2-alpha (assuming a semantic
1.0.0 corresponds to actual release
14.0.0). This gives developers (and tools like package managers) a straight forward way to update dependencies while minimizing the chance of breaking something.
Thanks again to everyone who has contributed in any shape or form. We’re so thankful for the help. Let’s try and make releases easier for everyone.