On Apr 23, 2012, at 11:19 AM, Robert Muir wrote: > On Mon, Apr 23, 2012 at 10:37 AM, Smiley, David W. <[email protected]> wrote: >> I understand your point Rob. But given your point of view, wouldn't it be >> sufficient for the README to simply say "download Ant 1.7 and…" without >> explicitly telling the user *not* to use Ant 1.8, which is suggestive of a >> particular problem with the newer version? You didn't tell them *not* to >> use Ant 1.9 yet either ;-) or… etc. It goes without saying when I read any >> instructions that references a particular version that that particular >> version works, and that there is no guarantee of anything else. >> > > I think what i meant is users usually see a required version and think > its a >= relationship.
Oh? But how could any software that depends on anything make such a claim? And we don't make that claim and anyone making such an assumption clearly has a flawed assumption. It's one thing to think, "the newer version will *probably* work", and such a thought might be characterized as an assumption -- we likely all think that. But if it fails, well, look back at what the README says to use. > So when users try newer versions of junit (and > all their tests failed), or things don't work quite right in ant, then > they think the whole project is broken. Aside from Java & Ant, and assuming the build is our Ant build (not an unsupported other build, maven or an IDE) I am unaware of how a user could _unwittingly_ use an incompatible library. They could bring it upon themselves by changing the build of course but then if the build fails, well clearly that is the first thing to check. >> Is there anything preventing Ant 1.8 being the release we advise people to >> use? After all, Ant 1.8 was already on my Mac after I had grabbed dev tools >> extras; I'm not sure when I last explicitly installed it. > > Yes: the fact we don't test it. That's not what I meant. I meant why don't we test against Ant 1.8 and make *that* the version we recommend? It is the latest Ant release, after all. ~ David --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
