|
From: Keith M. <kei...@us...> - 2007-10-18 19:30:38
|
On Wed, 2007-10-17 at 17:48 -0400, NightStrike wrote: > I'm sure that works, but why distribute an old version at all? A > release is a release, after all... OTOH, the distinction between a release candidate and a current release is rather arbitrary, especially in cases such as ours, where there is no shrink wrapped product. By definition, *any* release candidate has the potential become a current release; the transition occurs when the maintainer decides that no further patches will be applied against the most recent release candidate, in a particular branch. For GCC-3.4.5, that happened when Danny Smith declared that he would do no further work on that branch, and I relabelled the distribution packages accordingly; I also invited the maintainer of the installer to sync his package, but so far that has not happened. The release of MinGW-5.1.3 predates this promotion of some installable components from `candidate' to `current', so the .ini file bundled with the installer is obsolete, and the version to be dynamically downloaded at installer runtime still needs to be updated. IIRC, the bundled .ini file is provided as a convenience to users who are behind a proxy which blocks the dynamic download feature. However, I do see your point, that distributing a bundled obsolete version may lead to confusion. The problem is that, with different maintainers for the various packages, it is difficult to keep the installer in sync, especially when these various maintainers have other demands on their time; when you rely on voluntary effort, it is hardly reasonable to expect the volunteers to work to deadlines. Regards, Keith. |