Changes between Initial Version and Version 1 of Ticket #19016, comment 11


Ignore:
Timestamp:
08/12/15 18:02:41 (5 years ago)
Author:
nbruin
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #19016, comment 11

    initial v1  
    44"Change nothing". I think it's unfortunate that hashing by repr will usually work quite well in cases where a hash is possible: a hash usually requires a canonical form to compute the hash from and that canonical form will usually be used for printing, because that's nice. So in practice, the default hash (although much too slow to be a reasonable hash) will probably work when it should.
    55
    6 In cases where the current default hash *doesn't* work, I think that having it as it is now will lead to better detection than when we change it to a default hash. When people find this in particular cases, the solution is clear: actively disable the hash.
     6In cases where the current default hash *doesn't* work, I think that having it as it is now will lead to better detection than when we change it to a default hash of 0. When people find this in particular cases, the solution is clear: actively disable the hash.
    77
    88Having to *disable* something is of course the wrong way around: the default hash shouldn't be there in the first place. But putting a constant 0 hash there instead doesn't make transitioning to the desired situation any easier.