On Sat, Aug 31, 2019 at 06:56 PM, I wrote:
I am now writing the report; it should be finished later today / this evening UK time.
The original query (which I have to say I did not find entirely clear in its description of the problem) appears to have conflated two completely separate “phenomena” in the way Groups.io behaves. I would delete the word “anomalies” in my earlier post and replace it with phenomena for the simple reason that I don’t think either of the two can be described as faults, but may not qualify as full features either, although one of them comes rather close.
In all fairness the fact that two entirely separate factors are in play did not become really clear until I started investigating, and as far as I have been able to determine there is no interaction between them; neither modifies the action of the other, although on occasions it may seem that they do.
As might be realised from the original post the phenomena are exhibited by the behaviour of Hashtags in “unusual” circumstances (i.e. retrospectively applied) and the behaviour of Threading.
It is easier to consider threading first. A topic may be started from either the web UI or by email, and replies to that topic can be also sent either way. If the replies are sent via the web UI then by definition the messages will thread correctly, and this will still work after a long period of topic dormancy. (> 30 days) If the replies are sent by email then they will thread correctly, but at the time of writing it has not been possible to test if there is a “30 day rule” applicable to them for the simple reason that my usual modus operandum is by web so I have no “old” emailed posts to which I can reply. (I hope to check this in 31 days from now.)
There is, however, a second way of replying by email, and that is by composing a “new” email and typing Re: + subject in the subject line. This will thread correctly with the original topic provided that it has not been dormant for > 30 days; if the inactivity on that topic is longer than that then the “Re:” will be stripped off by Groups.io and the subject will appear as a new topic.
It must be noted that this method will fail if there is a typing error in the subject so that there is no match between an original subject and the reply in question, resulting in a new topic with a typing error starting; the Re: will be stripped off automatically. (As an aside the system seems tolerant of 2 spaces rather than one after Re: but the full extent of this tolerance has not been tested.)
Now… hashtags. Consider first a topic started with a hashtag applied; it matters not whether the topic was started by web UI or by email. The email sent to subscribers by Groups.io will have the hashtag in its subject line, and that hashtag will be perpetuated by the replies to the original message irrespective of whether those replies are by email or web UI unless for some reason a subscriber sends an email with a typed subject line that does not include the relevant hashtag. In this case the post will appear as normal as part of the topic but the emails sent out by Groups.io will not have the hashtag in the subject line either. All replies generated via the web UI will have the hashtag added.
If the hashtag is added retrospectively then the possibilities for confusion become greater. By way of example let us assume that a topic is started with no hashtags and that there are then 4 replies to it (5 posts in total). At that point a moderator edits the subject line and adds a hashtag. Following that, 5 subscribers post replies using the web UI, followed by another 5 who post via email.
Now the 5 posts done via the web will result in emails that have the hashtag in the subject line; the next 5 might be more complicated; if their (emailed) replies were replies to one of the initial 5 posts (prior to the addition of the hashtag) then their emails will not have the hashtag in the subject line and neither will the resultant distributed emails; if, however, any of them were replies to one of the messages sent from the web UI then they will have the hashtag in the subject line, and so will the resulting distributed emails. Over time – depending on how many posts the complete thread accumulates – some of the emails distributed to members will have hashtags in the subject line while others will not.
Now add the effects of threading and retrospectively – added hashtags together…
Looking back to the original query there are bits of evidence missing, e.g. the existence or otherwise of non – hashtagged threads that may have been split by virtue of the way various posts were generated; I think some more detailed investigation by the OP would be in order in the hope of getting a more comprehensive picture of what is happening on his group.
Now to a genuine oddity. A topic is started without a hashtag, and for some strange reason a subscriber (who may in fact be an owner or moderator) replies to it by email adding a hashtag to the subject line. (Don’t ask why they might do that!) That hashtag will not turn up against the topic in the web UI, but it will be included in distributed emails and will continue to propagate along the thread if anyone replies to a distributed email with the hashtag in the subject line.
My head hurts…