Sorry for a roundabout way of getting to the point, but wouldn't the characters used in matchers be treated in a special way, thus restricting their use in version strings?
On a related theme, should Ivy support semver ranges [1]? (that would include >, <, >=, <=, =, ||, &, !, and shorthands [2] *, ^, ~, " - ") There seems to be at least one Java implementation [3], but it might be too opinionated about version strings :-) [1] https://github.com/npm/node-semver [2] https://nodesource.com/blog/semver-tilde-and-caret [3] https://github.com/zafarkhaja/jsemver 2017-05-25 12:16 GMT+02:00 Nicolas Lalevée <nicolas.lale...@hibnet.org>: > Almost everything in Ivy is configurable, like the version matchers [1]. > > Ivy provide some sane default, but if some has a weird use case, he will > be able to tweak thing to make it work. > > Nicolas > > [1] https://ant.apache.org/ivy/history/latest-milestone/ > settings/version-matchers.html <https://ant.apache.org/ivy/ > history/latest-milestone/settings/version-matchers.html> > > > > Le 24 mai 2017 à 10:22, Gintautas Grigelionis <g.grigelio...@gmail.com> > a écrit : > > > > Sorry, I meant to ask what may happen if special characters from range > > specification are used randomly in a version string? > > > > Gintas > > > > 2017-05-24 10:15 GMT+02:00 Gintautas Grigelionis < > g.grigelio...@gmail.com>: > > > >> I was thinking about the discussion which touched upon restricted > >> characters in version strings. While Ivy does not place any > restrictions on > >> those strings, Ivy does support versions ranges as specified by OSGi. > So, > >> what happens if version range is used as a version string? > >> > >> Gintas > >> > >
