[Rebuild] Community feedback required - significance of word reversal in names
dermotm at gmail.com
Mon May 7 22:45:32 BST 2012
Over the long weekend I've done some hacking on the remaining failing
test cases. I concentrated on the quite-a-few cases where the tests
expect certain equivalent stings (mostly due to abbreviation) to be
treated as insignificantly different to each other. I've been able to
fix many of the failures and, subject to some conferring with Matt, I
think I can fix the test too.
There is one single test case that has been placed in the "needs clarity" file:
Its presence here indicates uncertainty that a community consensus
exists that the behaviour required by the test is correct. Having
looked at the test, I'm pretty sure I can make it pass, but it's best
to obtain community consensus first. The case is simple enough and
doesn't require ruby skills to understand. The test case takes an
instance of a Russian language street name, first added by an agreeing
mapper, then the words reversed by a non-agreeing mapper:
Original name: ул. Гая
Reversed version: Гая ул.
Aside: both version are abbreviated.
It turns out that "ул." is Russian for "street" (or similar), and that
there is no strong convention in Russian concerning the better order
in which to write names of this sort. The test as written contends
that the change made by the non-agreeing mapper is not significant.
I tend to agree that the change isn't significant. Likewise I'm not
aware of other languages in which it would be a significant change to
reverse the word order of a 2-word string.
My proposed measure to fix this test is to allow 2-word strings that
differ in only word order (allowing for equivalence of abbreviated
versions etc.) to be considered not significantly changed. I request
your opinions to help me decide whether this is a reasonable thing to
do (either for this specific example or for other similar cases). _If_
you believe that this case is safe but others not, please identify a
counter-example (this can itself, if the community agrees, be turned
into a test).
I will raise this case at LWG tomorrow, but for cases like this LWG
always prefers to go with community consensus. So let's have some!
Igaühel on siin oma laul
ja ma oma ei leiagi üles
More information about the Rebuild