[MlMt] Improving search performance?

[Glenn Parker]

On 25 Jun 2023, at 10:59, Robert M. Münch wrote:

> Using a classical database with indexes and the mail information MM 
> recognizes put into columns, should give mostly instant answers.

The validity of this assertion will depend on a lot of factors, 
including the size and content of the email corpus, along with the 
capacity of the machine performing the search.

I think the key point is that MailMate was not designed, much less 
optimized, to be a database search engine, because it doesn’t *need* 
to be one. Implementing that might easily require more of my machine 
resources than I am prepared to dedicate to an email client. Instead, it 
is, first and foremost, an email client that fetches, organizes and 
displays email efficiently. MailMate’s markup composition system is, 
frankly, a bit of a quirk, but it’s also an obvious tradeoff to make 
for a one-man dev team producing an email client. MailMate’s design is 
focused on what matters the most.

The search is fine IMHO, but it is also secondary. It supports 
infrequent, ad hoc queries with some highly-specialized filters on a 
wide range of email content, and it just works. The search performance 
is entirely adequate, especially when the scope of the search can be 
suitably limited. I rarely feel the need to search my entire email 
archive because I can almost always narrow down the scope, at least to a 
particular sub-folder based on the source or the year the message was 
sent. Those times when I do give up and search across everything, it’s 
usually so I can find my bearings when browsing older stuff. Then I can 
narrow down future searches based on the global search results. I admit 
that’s entirely my personal style, but it works for me.

Glenn P. Parker
glenn.parker at comcast.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freron.com/pipermail/mailmate/attachments/20230626/5302613c/attachment.htm>