Seq1 migration issue? #931
Replies: 1 comment 1 reply
-
I think this is a fair point. What I might do is just remove the
I think a migration approach would be simpler to be honest. If I migrate the conversion functions now (to |
Beta Was this translation helpful? Give feedback.
-
Release https://github.com/louthy/language-ext/releases/tag/v4.0.2 changed static constructor function
Seq1
toSeq
.I fully suport this decision as
Seq1
always felt like "wrong" or like a workaround.But I just want to point out some issue that might occur when upgrading existing code to this new release.
IMO a good way to migrate is via compile time errors (maximum safety).
Of course there are other options like
[Obsolete]
which might be helpful.I guess there are relevant use cases out there that might not result in compile time errors. This might produce bugs if users don't take special care (i.e. don't read release notes or maybe even ignore warnings about obsolete functions).
Here's some test that will fail on 4.0.2 (but not 3.4.15).
One idea for a safe migration path: Create some 3.4.16 (stable) release and another 3.x beta release where single parameter
Seq(x)
function is missing completely (forcing migration totoSeq
when updating to this intermediate version).There are other options using
[Obsolete]
and/or slow migration releases but IMO they are not appropriate for an OSS library like this (and maybe not safe).Beta Was this translation helpful? Give feedback.
All reactions