Any

Sorry for actually cloning the ticket, I didn't find another way to reopen it. Please see my comment:
https://support.zabbix.com/browse/ZBXNEXT-637?focusedCommentId=78622&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-78622
"I would say that this idea could be implemented in a different way: have separate triggers, have a chance to combine them to a single box in an "overview". The idea is to save some space on this screen. When you have lots of hosts and lots of triggers you do your best to get them all on one screen, and having 3+ similar triggers isn't a good approach." ©

For me the main wish is that I do not want to receive recovery message when situation become worse ("More than 90% used: OK (now 98% used)"). But I want to receive recovery notification when problem gone away (solved).
Then, alert comments should be preserved if problem changed severity.
And then, it will be good to have the same trigger-id for threading mail about changing status of the problem.

I think this can be implemented by some kind of grouping (linking) triggers. Additional option for dependency: inherit alert comments (and acknowledgement?) and do not send recovery message until any of these linked triggers is in alarm state.

As a large installation, we need to be able to activate different kinds of alerts to different groups of people depending on severity, which depends on value ranges.

Say a server rack is a bit too warm - a warning displayed on the dashboard is enough at first. Should the temperature rise more, DC people need to be emailed (average), then an SMS should be sent (high). Disaster measures follow. Etc, etc, etc.

Trying to get by with multiple/dependent triggers, there will be stacking alerts and/or those false OK's for things getting worse, impossibly complicated trigger expressions, an insane amount of triggers and so on.

One easy example for this ticket: You get an alarm for >5% free disk, but no emergency alert for =0%!
Its very confusing with triggers for every severity, but a grouping function could be helpful.

One workaround is to create several triggers withe the same name like this :

Name : Free disk space on volume

,pfree].last(0)}<5

Name : Free disk space on volume

,pfree].last(0)}<20 and {Your Template :vfs.fs.size[

,pfree].last()}>10
...
And on trigger overview for example show only Problems or Recent problems.