Big question that determines my opinion on all of this: would our current style of web interface still be available? My mailer doesn't like to give me plain text input, so when I need to see it I use the site. I'm also just used to the way things are, and don't want change. :)
-Aris On Thu, Jun 22, 2017 at 8:20 PM, omd <c.ome...@gmail.com> wrote: > So, almost exactly a year ago I said I wanted to transition the lists > to either GroupServer or Mailman 3, mainly for the sake of a better > web interface that includes the ability to post online (to reduce > friction for new users who might not want to sign up for 3 mailing > lists just to dip their toes in). ...And then I got anxious and > disappeared from Agora altogether. > > Well, I'm back and I thought it time to look into that again. > > First I tried setting up GroupServer. The code felt kind of bloated - > took like 10 minutes to install, mostly just to deal with the massive > number of Python packages it's split into. (And since I was trying to > use Docker, I had to wait for that install over and over. I gave up > on Docker.) Still, I eventually got it to work, and, yep, the web > interface is reasonably pretty… > > But then I tried sending a test email, and noticed that the email I > got had been wrapped in a garish HTML header/footer. Then I tried > sending a long message, and noticed that the email was cut off - very > early, in fact, after only 10 lines of text. Combining those two, the > impression I got was more of a forum notification than a real mailing > list post. I didn't see any option to change either behavior, in > either the (minimal) per-user settings or the administration settings. > > I could have tried to hack the code to just send mail > plaintext+unmolested, but the aforementioned bloatedness and confusing > setup made me leery of messing with the code, and I still had an > alternative. > > So today I installed Mailman 3 (+ HyperKitty + Postorius), which > predictably was a lot more annoying than it should be, partly because > it's split into three projects for no good reason. But so far, > impressions are better than GroupServer. I like the Web UI slightly > less, but it's not bad, and it does allow sending messages from the > browser - plaintext, mercifully. > > You can see it here: > > https://list.agoranomic.org/ > > The installation is currently entirely separate from the existing > Mailman 2 list, but I did import a copy of the list archives to > Hyperkitty so anyone interested can get a better feel for the UI. > Don't try posting there, but feel free to join and post on > agora-mm3test. > > I'd like to change the HTML so it shares the site header (and maybe > other styling) from https://agoranomic.org. > > The eventual plan would be to migrate the existing lists to Mailman 3. > Existing subscriptions and subscription settings would be brought > over; the base mbox archive would be continued from the existing > lists; all existing messages would be added to the HyperKitty site; > I'd also make a static copy of the existing Mailman 2 web archive, so > archive links would continue to work, although that would be separate > from the new site. In short, minimal disruption. > > Thoughts? Concerns? Suggestions? > > Known issues: > - The Hyperkitty importer is less than perfect. It doesn't like > messages with no Message-Id, or with non-ascii sender addresses, of > which we apparently have a bunch. Also, messages without a normally > formatted Date header get marked as new, including a report ais523 > sent in 2009 with the delightful "Date: some time near the end of > July", and some messages from Magu in 2003 with no Date at all. > - The test list adds a signature; it shouldn't. Also need to make > sure the subject line munging works the same way as currently. > - I should switch to something other than sqlite. > - The initial indexing takes a rather long time in Xapian and produces > a huge index compared to the input size. Maybe I should try switching > the backend to Solr or something. > - Hyperkitty is kind of slow. I tried some things to make it faster > already, but not enough I guess...
