Re: qualified imports, PVP and so on (Was: add new Data.Bits.Bits(bitZero) method)

On Tue, Feb 25, 2014 at 10:16 AM, Brandon Allbery wrote:

On Tue, Feb 25, 2014 at 2:21 AM, Sven Panne wrote:

...

Regarding upper bounds: I never understood what their advantage should be, IMHO they only lead to a version hell where you can't compile your stuff anymore *only* because of the bounds, not because of anything else.

A couple months ago we had yet another example of "that will never happen" caused by people ignoring upper bounds. Developers never saw any problem, of course; and who cares about all the users who had compiles explode with unexpected errors? I think it took less than two weeks after someone patched up the most visibly affected packages before developers were shouting to remove upper bounds from the PVP again, because the affected users are just users and apparently not important enough to consider when setting versioning policy.

I tried living without upper bounds. My attempt was not motivated out of disdain for users, but from the fact that all of the complaints I had had had been about the opposite, constraints that were too tight. The change was motivate largely by a desire to *improve* end user experience. However, after removing the bounds, the situations users wound up in were very hard to fix. From a POSIWID perspective, the purpose of removing upper bounds is to make Haskell nigh unusable without sandboxing or --force. Consequently, I reverted to PVP compliance within a month. Yes, compliance. Despite Henning's attempt to grab the moral high ground there, the PVP does not *require* the use of qualified imports to depend on minor versions, it merely indicates that doing so is a slight risk. To minimize breakage when new package versions are released, you can use

dependencies that are insensitive to minor version changes (e.g. foo >= 1.2.1 && < 1.3). However, note that this approach is slightly risky: when a package exports more things than before, there is a chance that your code will fail to compile due to new name-clash errors. The risk from new name clashes may be small, but you are on the safe side if you import identifiers explicitly or using qualificationhttp://www.haskell.org/haskellwiki/Import_modules_properly .

-Edward