1 TODO items for public-inbox
3 (Not in any particular order, and
4 performance, ease-of-setup, installation, maintainability, etc
5 all need to be considered for everything we introduce)
7 * general performance improvements, but without relying on
8 XS or pre-built modules any more than we currently do.
9 (Optional Inline::C and user-compiled re2c acceptable)
11 * mailmap support (same as git) for remapping expired email addresses
13 * support remapping of expired URLs similar to mailmap
14 (coordinate with git.git with this?)
16 * POP3 server, since some webmail providers support external POP3:
17 https://public-inbox.org/meta/20160411034104.GA7817@dcvr.yhbt.net/
18 Perhaps make this depend solely the NNTP server and work as a proxy.
19 Meaning users can run this without needing a full copy of the
20 archives in git repositories.
22 * HTTP, IMAP and NNTP proxy support. Allow us to be a frontend for
23 firewalled off (or Tor-exclusive) instances. The use case is
24 for offering a publicly accessible IP with a cheap VPS,
25 yet storing large amounts of data on computers without a
26 public IP behind a home Internet connection.
28 * support HTTP(S) CONNECT proxying to NNTP for users with
31 * DHT (distributed hash table) for mapping Message-IDs to various
32 archive locations to avoid SPOF.
34 * optional Cache::FastMmap support so production deployments won't
35 need Varnish (Varnish doesn't protect NNTP or IMAP, either)
37 * dogfood and take advantage of new kernel APIs (while maintaining
38 portability to older Linux, free BSDs and maybe Hurd).
40 * dogfood latest Xapian, Perl5, SQLite, git and various modules to
41 ensure things continue working as they should (or more better)
42 while retaining compatibility with old versions.
44 * Support more of RFC 3977 (NNTP)
45 Is there anything left for read-only support?
47 * Combined "super server" for NNTP/HTTP/POP3/IMAP to reduce memory,
48 process, and FD overhead
50 * Configurable linkification for per-inbox shorthands:
51 "$gmane/123456" could be configured to expand to the
52 appropriate link pointing to the gmane.io list archives,
53 likewise "[Bug #123456]" could be configured to expand to
54 point to some project's bug tracker at http://example.com/bug/123456
56 * configurable synonym and spelling support in Xapian
58 * Support optional "HTTPS Everywhere" for mapping old HTTP to HTTPS
59 links if (and only if) the user wants to use HTTPS. We may also
60 be able to configure redirects for expired URLs.
62 Note: message bodies rendered as HTML themselves must NOT change,
63 the links should point to an anchor tag within the same page,
64 instead; giving the user options.
66 * configurable constants (index limits, search results)
68 * handle messages with multiple Message-IDs (done for v2, doable for v1)
70 * handle broken double-bracketed References properly (maybe)
71 and totally broken Message-IDs
73 cf. https://public-inbox.org/git/20160814012706.GA18784@starla/
75 * improve documentation
77 * linkify thread skeletons better
78 https://public-inbox.org/git/6E3699DEA672430CAEA6DEFEDE6918F4@PhilipOakley/
80 * Further lower mail parser memory usage. We still slurp entire
81 message bodies into memory and incur 2-3x overhead on
82 multipart messages. Inline::C (and maybe gmime) could work.
84 * use REQUEST_URI properly for CGI / mod_perl2 compatibility
85 with Message-IDs which include '%' (done?)
87 * better test cases, make faster by reusing more setup
90 * large mbox/Maildir/MH/NNTP spool import (in lei, but not
91 for public-facing inboxes)
93 * MH import support (read-only, at least)
95 * Read-only WebDAV interface to the git repo so it can be mounted
96 via davfs2 or fusedav to avoid full clones.
97 davfs2 needs Range: request support for this to be feasible:
98 https://savannah.nongnu.org/bugs/?33259
99 https://savannah.nongnu.org/support/?107649
101 * Contribute something like IMAP IDLE for "git fetch".
102 Inboxes (and any git repos) can be kept up-to-date without
105 * Improve bundle support in git to make it cheaper to host/clone
106 with dumb HTTP(S) servers.
108 * Expose targeted reindexing of individual messages.
109 Sometimes an indexing bug only affects a handful of messages,
110 so it's not worth the trouble of doing a full reindex.
112 * code repository integration (cgit: done, TODO: gitweb, etc...)
114 * migration path to v2 without breaking v1 "git fetch" cronjobs
116 * imperfect scraper importers for obfuscated list archives
117 (e.g. obfuscated Mailman stuff, Google Groups, etc...)
119 * improve performance and avoid head-of-line blocking on slow storage
120 (done for most git blob retrievals, Xapian needs work)
122 * HTTP(S) search API (likely JMAP, but GraphQL could be an option)
123 It should support git-specific prefixes (dfpre:, dfpost:, dfn:, etc)
124 as extensions. If JMAP, it should have HTTP(S) analogues to
125 various IMAP extensions.
127 * search across multiple inboxes, or admin-definable groups of inboxes
129 This will require a new detached Xapian index that can be used in
130 parallel with existing per-inbox indices. Using ->add_database
131 with hundreds of shards is unusable in current Xapian as of
132 August 2020 (acknowledged by Xapian upstream).
134 * scalability to tens/hundreds of thousands of inboxes
136 - pagination for WwwListing
138 - inotify-based manifest.js.gz updates
142 * lei - see %CMD in lib/PublicInbox/LEI.pm
144 * make "git cat-file --batch" detect unlinked packfiles so we don't
145 have to restart processes (very long-term)
147 * support searching based on `git-patch-id --stable` to improve
148 bidirectional mapping of commits <=> emails
150 * linter to check validity of config file
152 * linter option and WWW endpoint to graph relationships and flows
153 between inboxes, addresses, Maildirs, coderepos, newsgroups,
154 IMAP mailboxes, etc...
156 * pygments support - via Python script similar to `git cat-file --batch'
157 to avoid startup penalty. pygments.rb (Ruby) can be inspiration, too.
159 * highlighting + linkification for "git format-patch --interdiff" output
161 * highlighting for "git format-patch --range-diff" output
162 (linkification is too expensive, as it requires mirroring)
164 * support UUCP addresses for legacy archives
166 * decode (skip indexing of) base-85 binary patches to avoid false-positives