Codebase list fdm / run/9b4373ac-c431-4768-90fa-95dfd1ca4dc1/main TODO
run/9b4373ac-c431-4768-90fa-95dfd1ca4dc1/main

Tree @run/9b4373ac-c431-4768-90fa-95dfd1ca4dc1/main (Download .tar.gz)

TODO @run/9b4373ac-c431-4768-90fa-95dfd1ca4dc1/mainraw · history · blame

- SMTP AUTH, SSL and/or TLS
- auth on http proxying
- http (hotmail) fetching
- redo chrooting properly (resolver needs to be in parent)
- match attachment expressions:
	match attachment (index > 1 and size < 200 and type "fnmatch")
- attach get name from content-disposition
- more attachment stuff: remove-attachment, write-/pipe-attachment etc
- refactor & clean up privsep mail transfer code
- is there any reason struct mail shouldn't point to a struct account?
- cumulative effect of a compound action sucks
- better website
- fetch from compressed mbox: have to change fetch-mbox to not use mmap
- all timeouts should be in seconds or milliseconds, decide which and make it so
- use TOP to grab only header and return to get body only if actually necessary
- automatically delete old mail on server once it reaches a certain age...
  does this break the barrier of what fdm is designed for?
- does imap need a to-fetch list a la pop3? having mails marked read that have
  not been successfully delivered is still a problem though. how can this be
  fixed?
- string match as regexp, attachment as fnmatch sucks
- common comparison code? num/str
- pop3 over pipe lack of .netrc sucks
- generic flags on account so that no-verify/no-apop don't have to be in the
  right order also fix man page
- NNTP should warn if mail is dropped - generic flag
- make-style filtering for tags :S/// etc
- docs from Thomas@BIC:
	clarify that first match stops without continue
	shell commands: when evaluated (done?)
	clarify keep for NNTP (done?)
	clarify action tag use/updating, and wrt lambda actions
	man page: "The name of the last action executed for this mail."
		(last or current?)
	attach size units (same as maximum-size)
	case sensitivity
	mention short-circuit evaluation
	check pipe command user
- when doing .netrc lookup, match the username if given as well to allow
  multiple users on the same server
- set secondary groups when changing user
- regress fails as root
- rml could be stored as actual tags, then command0-9 could go... means
  copying... get rid of %0
- add FAQ entry about no-apop since there are several servers out there
  which claim to support it but do not
- option to relax group ownership checks to permit secondary groups,
  for shared mailboxes
- examples should have more explanation
- some sort of against-file or matches-file command to match against
  a whitelist. bloat vs grep?
- privsep/child error handling is crappy. audit. how to do this better?
- NNTP over ssh/pipe