Hi,
I am not sure its intended or a fault in the newest Dovecot versions. I have been using Dovecot v1.2.15 on Debian squeeze and FTS is working as expected. When I search a quoted string "very good", I get 107 results including plain and HTML emails which have this phrase.
In order to compare the benefits of lucene over squat, I recently started testing dovecot v2.2.13 on Debian Sid with the same maildir content. But now the same search "very good" yielded just 8 results. I thought it could be some problem with lucene so I tried switching to squat and got 107 results again. After this I deleted the old squat search index files created by v1.2.15 and re-indexed the mail-box by using doveadm index command. Now the same squat search is giving 8 results just as lucene. So I have realized that its not a problem with just lucene but FTS in newer dovecot isn't indexing those emails which have Content-type as text/html.
Thus if a mail is like this:
Content-Type: text/html
<b>He is very good.</b>
It isn't shown in search by the squat indexes created using dovecot v2.2.13. I have done further testing on some sample emails which confirmed this behavior.
Why is this so?
-Regards, Akash