[
https://issues.apache.org/jira/browse/MAILBOX-200?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13819940#comment-13819940
]
Robin Bankhead edited comment on MAILBOX-200 at 11/12/13 11:32 AM:
-------------------------------------------------------------------
OK, I will run those as soon as I have a moment to figure out how.
>From yesterday's log-diving I learned that LSUB works OK (as it should, the
>subscriptions file is maintained fine) but LIST "" * returns no subdirs.
>Thunderbird accepts what LSUB tells it, Horde verifies with LIST so it
>doesn't acknowledge the folders' presence.
Using netcat, SELECT appeared to work inconsistently also: first it returned
BAD on a subdir, but if it was first called on INBOX, the previous SELECT then
did succeed on the original subdir.
It might still be simple enough that the only significant refactoring to be
done is in MaildirMailboxMapper.findMailboxWithPathLike(), but certainly I can
say the initial approach was wrong. I could of course keep the delimiter code
in, but primarily for the reasons its stub implementation was there to begin
with (it being noted as implementation-specific in the spec), not really for
this.
was (Author: headbank):
OK, I will run those as soon as I have a moment to figure out how.
>From yesterday's log-diving I learned that LSUB works OK (as it should, the
>subscriptions file is maintained fine) but LIST "" * returns no subdirs.
>Thunderbird accepts what LSUB tells it, Horde verifies with LIST so it
>doesn't acknowledge the folders' presence.
Using netcat, SELECT appeared to work inconsistently also: first it returned
BAD on a subdir, but if it was first called on INBOX, the previous SELECT then
did succeed on the original subdir.
It might still be simple enough that the only significant refactoring to be
done is in MaildirStore.getFoldersWithNameLike(), but certainly I can say the
initial approach was wrong. I could of course keep the delimiter code in, but
primarily for the reasons its stub implementation was there to begin with (it
being noted as implementation-specific in the spec), not for this.
> Hierarchical folder layout for maildir
> --------------------------------------
>
> Key: MAILBOX-200
> URL: https://issues.apache.org/jira/browse/MAILBOX-200
> Project: James Mailbox
> Issue Type: Improvement
> Components: maildir
> Affects Versions: 0.6
> Environment: Windows 7 Professional 32bit
> Reporter: Robin Bankhead
> Priority: Minor
> Attachments: mailbox-maildir-hierarchical-v3.patch
>
>
> Propose adding support in maildir for hierarchical on-disk layout of folders,
> as used by some other MTA/MUAs (Dovecot for example has optional support[1]).
> Secondary-, tertiary- etc level subfolders would be represented by an
> equivalent dir/subdir structure in the filesystem, so e.g.
> [maildir].Dir.SubDir
> would exist on-disk as
> [maildir]/.Dir/SubDir
> The motivation is simply to have on-disk layout more closely mirror that of
> the mailbox folder structure, though it has the additional fringe benefit of
> relieving the pressure imposed by subdirs-per-dir limits in some filesystems
> (though it would take a *very* large mail corpus to be affected by that, I
> admit).
> This simple proof-of-concept patch to MaildirStore.java enables the feature.
> If this is of interest, I would look to make it a configurable option in
> mailbox-maildir.properties (which I don't currently know how to do).
> [1] http://wiki.dovecot.org/MailboxFormat/Maildir#Directory_Structure
--
This message was sent by Atlassian JIRA
(v6.1#6144)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
