[BALUG-Admin] Status/update on list migrations

Michael Paoli Michael.Paoli@cal.berkeley.edu
Thu Jul 27 06:24:21 PDT 2017


So, quick(ish) update.

Getting closer to first actual migration - and progress
is certainly more than "inching" forward.

By yesterday A.M., had whipped up bit of code aimed towards
web-scraping all member preferences (except their password).
Could ask DreamHost for data dump of that, but the work involved
to get them to actually do that times the probability they
wouldn't do it anyway or would screw it up - I much prefer
going with the sure-fire web scrape - and yes, that'll mean
new passwords for members, but if that's about all they have
to deal with, I figure that's pretty good.  And I much prefer
to not have well over 800 members all have their list preference
settings all get squashed to default or the same across all on
per-list basis - that wouldn't be very friendly.
Anyway, code isn't completed, but got it yesterday A.M. through
point where it gets all the member's email addresses, and has the
URLs to each member's individual preference page - which has all their
settings except can't extract existing password.  So, bit more
coding to grab/parse that, and fairly close to migration ready.

The other bit is *loading* preferences.  Have looked over Mailman
DB stuff - it uses Python pickle format.  I don't (yet) grok
Python, but looks highly doable for me to work that out.
Looks like all the per-user settings are basically some
strings (e.g. optional full name if entered),
and a binary mapping (e.g. 2^1 bit is toggle for option 1,
2^2 bit is toggle for option 2, etc. (at least approximately -
might be some with more than 1 bit for options with more than
binary selection).  Anyway, shouldn't be too hard to work out
those various bits and test (probably do some add/drop of a
scratch/test list for testing some of that out - also at same
time, tweak Mailman config a bit so any newly created list
comes into existence with desired defaults).

Hmmm, I should'a written the web scrape stuff to pull roster years
ago ... would'a saved me about 5 to 10 or so minutes a month
(okay, still would'a been fairly long ROI, but would've paid off
by now).




More information about the BALUG-Admin mailing list