[JB]
As for the suggestions about re-organziation to improve
customizability, you are fundamentally right. My design focus was
service (www.mail-archive.com) oriented as opposed to product
oriented. If the focus is product oriented, re-organization to support
customization would be very helpful.
I am not currently planning a re-organization of this sort, as doing a
good job, while not compromising simplicity, will be difficult and
time consuming. My energies are still service oriented. However,
don't let that stop you, and if you make clean changes, please send'em
in!

[AL]
Understood that your priority is service oriented, not developing
product.
However you have taken the first step of releasing code with reasonable
expectation that as well as helping others it may help your service from
feedback.

My view is that apart from benefit to others you will need to take the
second step of re-organization to support customization (though not much
more than re-organization) in order to get much benefit for your
service, precisely BECAUSE doing a good job while not compromising
simplicity will be difficult and time consuming. It will be less
difficult and time consuming for you because of your familiarity with
the code than for others so is likely to be delayed if left to others
and you are less likely to compromise simplicity than others would.

At present anybody implementing will be more inclined to just do the
customization they need for their own site without taking on the
re-organization job (or just customize MHonArc directly for a single
local site administered directly as usual rather than retaining your
approach of enabling remote users to add lists by email).

Once you have done the re-organization (without also taking further
steps to make customization as simple as possible since that is not your
priority), including mailme in the ports/packages collections for
standard Unixes like FreeBSD and various Linux distributions will enable
significant numbers of people to run local archive services based on
yours, as a result of which there would be a flow of patches,
suggestions and perhaps significant add-ons that really would benefit
your service (as well as further improvements in customization made
feasible by the re-organization).

Standard Linux RPM and FreeBSD pkg_add facilities with dependencies on
MHonArc, Ht://Dig, MH and any other "required" packages to be installed
first with "plain vanilla" customization could result in large numbers
running it, not as a public archive service like yours, but simply for
convenience in subscribing to mailing lists in the same way that you
browse your own service for the lists you are interested in following.

BTW, re-organization should take account of the possibility of switching
from shell to perl so that somebody could implement a package for
Windows platforms too. Since MHonArc is capable of running under Windows
there would be potential for even larger numbers of users (though a
smaller proportion would submit patches).

[JB]
As to your specific suggestions, I think keeping the rcfile intact (as
opposed to generated in pieces) might be important in keeping things
understandable. However, internal organization of that file and
embedded comments might be improved [i.e. organize the file into
separate sections (along the lines you mentioned?) and
documentinteractions]

[AL]
Understood. Perhaps there should be an rcfile.model with paramaters some
of which are also used in digger.model etc (and per list option -
generate $HOME/archive/$MAILLIST/rcfile from rcfile.model at list
subscription time instead of $HOME/rcfile at install time). My
suggestions were just off the top of my head. You know what needs to be
done. Even if you don't agree with above advice to do the
re-organization yourself perhaps you could write a plan for it spelling
out exactly what should be done and how, which might encourage others to
do it or might eventually be used by yourself (with possible benefit of
further suggestions from others before proceeding), but either way would
be more likely to avoid compromising the simplicity than if left to
others to design as well as code.

Reply via email to