On Wednesday 17 December 2008, Jerry James wrote:
On Tue, Dec 16, 2008 at 10:30 PM, Stephen J. Turnbull
> In fact we have two fixes. One which involves changes to 21.4 and
> thus is not backward compatible to existing installations in the sense
> that anyone with an existing installation < 21.4.22 who wants to build
> all packages would have to upgrade to 21.4.22. The other involves a
> monkey-patch which could be added to the package build process and
> used only on those installations that don't support it (it involves
> adding a new variable, and thus a boundp test for that variable would
> be definitive). The former fix is preferred by Mike but Vin isn't
> doing anything about it, and the latter requires monkey-patching,
> which has its own set of problems.
Is Vin not doing anything about it because he doesn't have time, or
because he disagrees with that direction?
How about a 3rd approach: simply revert (temporarily or move to a branch) or
otherwise get out of the way the packages/updates that cause the current
failures, and release the rest? And possibly rebuild everything without the
offending new/updated packages and let the resulting set cook in experimental
for a while before release.
XEmacs-Beta mailing list