statrs 0.6.0 and 0.7.0 beta regression

[est31]

statrs versions 0.6.0 and 0.7.0 regressed from stable to beta (full log 1, full log 2):

Aug 03 23:18:26.668 INFO blam! error[E0061]: this function takes 1 parameter but 0 parameters were supplied
Aug 03 23:18:26.668 INFO blam!  --> <anon>:7:11
Aug 03 23:18:26.668 INFO blam!   |
Aug 03 23:18:26.668 INFO blam! 7 | assert!(x.max().is_nan());
Aug 03 23:18:26.668 INFO blam!   |           ^^^ expected 1 parameter
Aug 03 23:18:26.668 INFO blam! 
Aug 03 23:18:26.668 INFO blam! error[E0599]: no method named `is_nan` found for type `[_; 0]` in the current scope
Aug 03 23:18:26.668 INFO blam!  --> <anon>:7:17
Aug 03 23:18:26.668 INFO blam!   |
Aug 03 23:18:26.668 INFO blam! 7 | assert!(x.max().is_nan());
Aug 03 23:18:26.668 INFO blam!   |                 ^^^^^^
Aug 03 23:18:26.668 INFO blam! 

cc @boxtown

[arielb1]

• #42496 - The functions Ord::min and Ord::max were added.

This conflicts with functions with the same name being implemented by other traits, e.g. by trait Statistics defined in https://github.com/boxtown/statrs/blob/v0.7.0/src/statistics/statistics.rs

To fix this, either rename the conflicting functions, or call them using UFCS notation (Statistics::max(x).is_nan()). This is unfortunate.

[boxtown]

Thanks for the heads up, I'll address this ASAP. It'll make it to the 0.8.0 release which I'm hoping to cut by Monday

…

On Sat, Aug 5, 2017 at 10:52 AM, Ariel Ben-Yehuda ***@***.***> wrote: - #42496 <#42496> - The functions Ord::min and Ord::max were added. This conflicts with functions with the same name being implemented by other traits, e.g. by trait Statistics defined in https://github.com/boxtown/statrs/blob/v0.7.0/src/statistics/statistics.rs To fix this, either rename the conflicting functions, or call them using UFCS notation (Statistics::max(x).is_nan()). This is unfortunate. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#43680 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ACwkvmJEWj-YzFE132NeOw-prqRRC6Nbks5sVIHJgaJpZM4Oudmc> .

[alexcrichton]

@boxtown note that the change here seems to have caused more breakage than intended, so if you've got difficulty fixing this please let us know!

[boxtown]

Regression fix is in place in master (see boxtown/statrs@e728aac) and will be released with 0.8.0. It was only a small regression affecting doc tests for empty slices and can be easily resolved by pinning the type of the slice to [f64; 0] e.g. let x: [f64; 0] = [];. No API or code changes were necessary

[alexcrichton]

Ok thanks for the info @boxtown!

[boxtown]

0.8.0 was just released yesterday so this issue can probably be closed

[est31]

Thanks!