are you still working on this?

As see in my fix is creating problems, so I had to revert it and I'm looking for another solution.

Please do not close. I need to follow-up with upstream and get a real fix done somehow.

I also wish to track how I'm going to maintain this workaround in the meanwhile, as for the moment it is manually applied and will vanish on the next code update.

looks like it's working fine, thanks !

Quack,

Upstream (Django) refused to consider the bug; I guess it was too difficult to get this function to work well with HTML content and they gave up. I dug into the problem and produced a patch which is not causing regressions according to the tests, and asked for reconsideration.

In the meanwhile I backported this fix and others related to this function in a special branch. I manually applied the patch on the ML server and it seems all is fine. could you have a look and tell me if you find any problem?

I'm not sure how to maintain this over time. I would not like to put the django package on hold and miss important security fixes. So I guess we'll have to reapply after each upgrade. Maybe I can add this to the playbook.

I did not know Gmail had this behavior.

I found the problem originates in Django and the MM3 developers are aware of it: https://gitlab.com/mailman/hyperkitty/issues/29

I added some more info and opened a BR on Django as the original fix did is not sufficient. I'll dig deeper and see if I can propose a fix. We won't get it in time for CentOS 8 so this means also creating a not too ugly fix. Happily the Hyperkitty code using this function is using a wrapper `urlizetrunc()` in `hyperkitty/templatetags/hk_generic.py` so it should be easy to fix this in a single place.

_o<

I’m not using brackets on purpose, that’s how gmail send the emails with urls

I confirm the problem but that's not something we can change in the settings. Sandro uses angle brackets to wrap URLs and despite the fact that's text/plain it seems to be interpreted or at least the URL boundary is missed. I wonder why it is not using the text/html alternative which is fine (I checked the raw data). So I'm going to look into it but that's probably going to take some time to read the code and find the source of the problem.

As for the "high priority" statement, this does not make me care more, or less, and that's somewhat exaggerated despite being quite annoying. The service works, and noone complained for months, people rarely use brackets around URLs in mails and there's an obvious workaround (resend without the brackets which I believe are not essential to the good reading of this news).

can you check?