On Wednesday, October 1, 2003, at 07:22 am, Gour wrote:
Richard Barrett (r.barrett@openinfo.co.uk) wrote:
I have uploaded updated files to sourceforge for the following patches and bugs I maintain, which include my Mailman-HTdig integration patches. You can reach them through:
Is there any chance for these patches to become integrated in the official distribution.
A perfectly reasonable question.
We are just configuring one server which runs Gentoo and it is a pain to always manually patch Mailman, instead of plain "emerge".
It is pain for me too, as the author and maintainer of these patches, particularly #444879 and #444884. I have to retest and, if necessary, rework and again retest them against each new MM release. Yes, I would have to do some of this work as I use the patches on systems I manage but the time and effort to try an deliver a trouble free and documented patch to a wider community arguably doubles the effort I expend.
As I write, the sourceforge page for patch #444884 appears to be broken (again), probably because the number of files associated with the patch has hit some PHP limit when rendering the page - that was the problem the last time this occurred.
All these files have accumulated to cater for the numerous releases of MM since I first developed the patch (and some fixing of my bugs introduced by my patched code). But I do not have the necessary privileges to delete the older, obsolete files from sourceforge.
But the real problem is that no consensus has been reached on committing Mailman to use of the HTdig search engine for archive search. With posts to mailman-users and mailman-developers lists by some subscribers that express negative views about HTdig, this consensus is not likely to be reached. Also, I believe the main MM developers are not keen on bundling HTdig with MM or tying MM too tightly to HTdig.
I have a half completed rework of the #444884 patch to make it more of a "search framework", so that a variety of search engines could be plugged in more readily, but finding the time and effort to complete this is a problem for me. And, if I am honest, I doubt that even this would be admitted to the core MM code before MM 3. Take patch #444879, which has been entirely search engine agnostic and configured to do nothing be default for some considerable time, and that has not been accepted into the CVS. To cap it all, MM 3 is supposed to get a better built-in archiver (than the current pipermail implementation) incorporating built-in search; so why waste my time and effort on figuring out enhanced changes to obsolescent code.
Others must speak to their view but I doubt you will see this problem resolved before MM 3.
Sincerely, Gour
-- Gour gour@mail.inet.hr Registered Linux User #278493
Richard Barrett http://www.openinfo.co.uk