Fixes #623. LockDiff now ignores hash/solvemeta.

[mikijov]

Fixes #623. LockDiff now uses only project dependencies to identify changes, thus ignoring any changes to the SolveMeta iformation. If required changes in the SolveMeta can still be seen using source diffs. Modified and added tests to check new behaviour.

The solution was suggested by @sdboyer in #623. The code change is simple. I am more looking for comments if you can think of any unintended consequences of not comparing hashes any more.

[sdboyer]

i'm looking this over and i think it looks good. i started in on the implications of not comparing the hashes, and i think there mostly aren't any, as we do all the necessary comparisons earlier on, prior to the txn writer (where this is actually used). that's most of the design.

however, i think we may need one manual check of the hashes, as even if nothing changes in the locked project set, we do still need to write out a new lock if the hash changes.

...hmmm. actually, i'm thinking about this, and i'm not sure about the UX now. i think i originally wanted this because i didn't want to bother the user when the input hash changes, but i'm not sure that's right. we always have to rewrite the lock with the new hash, and we should still probably tell the user that such a change occurred.

maybe i need to go back and revisit my original motivations. inputs/thoughts welcome 😄

[carolynvs]

In my mind, the purpose of the lockdiff is to support dry-runs, and giving people fair, accurate warning of what dep is about to do. As a user, if I ran dep ensure -n and dep reported back that it would do nothing, then when I ran it, it rewrote the lock, I would feel that the dry-run can't be trusted. 😀

[sdboyer]

@carolynvs yeah, i think you're right. this was a bad idea 😞 sorry, @mikijov.

closing.

[mikijov]

No problem. Please mark few more issues as help-wanted. :-)

[sdboyer]

yes! such a wonderful request. 🎉😀 I'm going to do a pass through soon, perhaps especially focusing on existing TODOs in the code.

…

On July 22, 2017 11:17:44 AM EDT, "Milutin Jovanović" ***@***.***> wrote: No problem. Please mark few more issues as help-wanted. :-) -- You are receiving this because you modified the open/close state. Reply to this email directly or view it on GitHub: #788 (comment)