Re: Item "255451" in collection "108" has no RID.

Dear Werner, dear community,

Werner Joss - 20.05.20, 17:08:46 CEST:
> Am Mittwoch, 20. Mai 2020, 13:52:06 CEST schrieb René J.V. Bertin:
> > >just a side note:
> > >I tried https://kube-project.com/ some time ago, which worked quite
> > >impressively then, without any akonadi* stuff whatsoever
> > 
> > When I tried it it had no kwallet integration, and it had a QML
> > interface. The former is resolved by now, hopefully, the latter
> > probably not - it's something you have to appreciate (the "speed",
> > and the mobile-style design.
> Yes, the style is more like mobile, that's true.
> Speed was ok for my use case.
> I'm currently still on kmail (5.14.1) which works quite ok here, with
> some glitches. I'll keep an eye on kube, though...

I think Sink has potential. But at the moment it does not even come 
close to feature parity with Akonadi. Does it need all its features? 
Very likely not. However… as I am still using POP3 and also if I switch 
to IMAP would prefer to move older mails to local storage instead of 
having years of mail history sit on my server, it currently is no option 
for me.

That said, while I defended Akonadi quite heavily in the past, meanwhile 
I think it may be good to replace it. Why?

This issue and other issues with Akonadi are there for a long time 
*despite* the enormous efforts of developers to fix it up. I remember the 
action plan of Dan to fix it, I have seen the huge efforts of Dan and 
other developers to make it speedier and more reliable. However… while 
there were definitely improvements, it still has huge reliability and 
performance issues, even when I clean out old mails via archivemail. I 
am aware that his last intended step to "Make Indexing Great Again" is 
still to be completed¹. And Dan if you are reading this: This is in no 
way a criticism about your activities. I know you did the best you could 
and that it is a lot of work and I am very grateful for what you have 
done so far.

Added to that a user of a mail application shall never ever need to care 
about database administration. Granted MySQL worked very, very well for 
a Zimbra server, but there the admin took care of it. And luckily I did 
not even need to, except for backup. It was just running nicely with 
abysmal performance settings. But it appears to me that Zimbra used the 
database in such an efficient way that it did not even matter, cause it 
was performing just fine with folders of almost 500000 – yes, you read 
that right – mails in them. But of course the web client tricked on 
them: They just showed the first 1000 or so and loaded more as I scrolled 
down. Fine with me however.

That all written, probably time for me to test out Kube myself. I bet it 
is blazingly fast.

[1] Make Indexing Great Again

https://phabricator.kde.org/T7014

Best,
-- 
Martin