sup, branch HEAD

update wiki submodule

2026-03-28T03:50:09Z

commit beff97f84382cbcf3494ab1fd87b7b076ee0a2ca parent 7580318c9d58be5a24b320c0d69a87a319292d2d Author: Dan Callaghan Date: Sat, 28 Mar 2026 14:50:09 +1100 update wiki submodule Deletes the obsolete sup-psych-ify-config-files(1) man page.

adjust regex for fixing old YAML tag URIs to handle SentLoader

2026-03-28T03:35:09Z

commit 7580318c9d58be5a24b320c0d69a87a319292d2d parent e80ea65987671aeedc0e493ba42f7c832e96870b Author: Dan Callaghan Date: Sat, 28 Mar 2026 14:35:09 +1100 adjust regex for fixing old YAML tag URIs to handle SentLoader Apparently some old sources.yaml files would have a source like this: - !supmua.org,2006-10-01/Redwood/SentLoader {} The regex previously added in commit 1aac9d4388 would fail to fix up an invalid tag URI like this because of the trailing empty map. Fixes #516.

delete top-level shell.nix symlink

2026-03-15T04:08:20Z

commit e80ea65987671aeedc0e493ba42f7c832e96870b parent c39852484ff5dc9680edab8e9b10b9ec568b3252 Author: Dan Callaghan Date: Sun, 15 Mar 2026 15:08:20 +1100 delete top-level shell.nix symlink I keep forgetting that it doesn't actually work, because nowadays nix tries to resolve the reference to gemset.nix relative to the symlink not its real path.

update links to point at the new supmua.dev domain

2025-10-12T00:47:48Z

commit c39852484ff5dc9680edab8e9b10b9ec568b3252 parent 4ab9d747ae5f52485a5393cac9435e1ac9202a83 Author: Dan Callaghan Date: Sun, 12 Oct 2025 11:47:48 +1100 update links to point at the new supmua.dev domain

refuse to start if default account is missing from config

2025-05-18T07:36:05Z

commit 4ab9d747ae5f52485a5393cac9435e1ac9202a83 parent 67e38b99c44d8d882d3f178095a5968a2ea54dd9 Author: Dan Callaghan Date: Sun, 18 May 2025 17:36:05 +1000 refuse to start if default account is missing from config Fixes #533.

refuse to start if sendmail command is not configured

2025-05-18T07:15:36Z

commit 67e38b99c44d8d882d3f178095a5968a2ea54dd9 parent 861becad89df65d55ed3cbd001ab36b909154ca0 Author: Dan Callaghan Date: Sun, 18 May 2025 17:15:36 +1000 refuse to start if sendmail command is not configured Fixes #524.

rescue TypeError when sending mails

2025-05-18T06:59:35Z

commit 861becad89df65d55ed3cbd001ab36b909154ca0 parent 0e1d70ece10efc171e5b91d101d253b1cc4db2a6 Author: Dan Callaghan Date: Sun, 18 May 2025 16:59:35 +1000 rescue TypeError when sending mails If the user has not set the "sendmail" key in their account config, we will try to invoke Popen on nil, which is a TypeError. Showing this TypeError to the user in a flash message is not very nice, but it's still better than crashing and losing their draft. Relates to #524.

ncursesw: 1.4.12 → 1.4.13

2025-05-18T04:51:29Z

commit 0e1d70ece10efc171e5b91d101d253b1cc4db2a6 parent a927293263d4c7c48bfa925d23a648ebca0d12db Author: Dan Callaghan Date: Sun, 18 May 2025 14:51:29 +1000 ncursesw: 1.4.12 → 1.4.13

assume draft charset is $encoding instead of ASCII when indexing

2025-05-11T06:10:16Z

commit a927293263d4c7c48bfa925d23a648ebca0d12db parent beeae7c6e260d3ea5b262fac2d1096c2c8858514 Author: Dan Callaghan Date: Sun, 11 May 2025 16:10:16 +1000 assume draft charset is $encoding instead of ASCII when indexing Sup writes drafts without any Content-Type header. It assumes when sending the mail that it is text/plain with charset=$encoding. Assume the same when indexing the draft. Fixes #343.

use $encoding for drafts instead of forcing UTF-8

2025-05-11T06:09:00Z

commit beeae7c6e260d3ea5b262fac2d1096c2c8858514 parent 39289820a7a9573d0ca830ce0cb2881deba87aa8 Author: Dan Callaghan Date: Sun, 11 May 2025 16:09:00 +1000 use $encoding for drafts instead of forcing UTF-8 If the user's locale charset is not UTF-8 (unlikely) let's respect that, instead of forcing drafts to always be treated as UTF-8.

ncursesw: 1.4.11 → 1.4.12

2025-05-04T07:55:08Z

commit 39289820a7a9573d0ca830ce0cb2881deba87aa8 parent ccfaadf2740009b6adeb8d70ee472c225ffe72bf Author: Dan Callaghan Date: Sun, 4 May 2025 17:55:08 +1000 ncursesw: 1.4.11 → 1.4.12

parse drafts as individual messages, not mboxes

2025-05-03T11:10:46Z

commit ccfaadf2740009b6adeb8d70ee472c225ffe72bf parent a0a496fa0fd6cbda3c9768bf0b8ad18b8f9013ce Author: Dan Callaghan Date: Sat, 3 May 2025 21:10:46 +1000 parse drafts as individual messages, not mboxes As far as I can see they were always being written with one message per file and no From envelope (maildir style) so there is no reason to try iterating the draft as if it were an mbox. Fixes the pathological case of trying to load an empty file as a draft.

handle drafts with missing Date header

2025-05-03T05:58:20Z

commit a0a496fa0fd6cbda3c9768bf0b8ad18b8f9013ce parent a2bd945dc883e8f65222e84c61a72bd3541fac70 Author: Dan Callaghan Date: Sat, 3 May 2025 15:58:20 +1000 handle drafts with missing Date header

fix typo in DraftManager#discard handling already deleted draft

2025-04-27T07:45:59Z

commit a2bd945dc883e8f65222e84c61a72bd3541fac70 parent aeefd695c6ebeb8c486590a49e38db466f43c3bc Author: Dan Callaghan Date: Sun, 27 Apr 2025 17:45:59 +1000 fix typo in DraftManager#discard handling already deleted draft

tests: add rudimentary coverage for drafts

2025-04-27T07:42:09Z

commit aeefd695c6ebeb8c486590a49e38db466f43c3bc parent efdca8c7195228698ccd6a4dc21debd478014e61 Author: Dan Callaghan Date: Sun, 27 Apr 2025 17:42:09 +1000 tests: add rudimentary coverage for drafts

prepare for next release

2025-04-21T04:37:33Z

commit efdca8c7195228698ccd6a4dc21debd478014e61 parent 8173db9f7bd3cbc6c322c5b375563347393103b2 Author: Dan Callaghan Date: Mon, 21 Apr 2025 14:37:33 +1000 prepare for next release

release 1.3

2025-04-21T04:34:35Z

commit 8173db9f7bd3cbc6c322c5b375563347393103b2 parent 5dc5846e6e86f31f3e890ccbd92d2df620755ac4 Author: Dan Callaghan Date: Mon, 21 Apr 2025 14:34:35 +1000 release 1.3

update wiki submodule

2025-04-21T03:56:59Z

commit 5dc5846e6e86f31f3e890ccbd92d2df620755ac4 parent 104e89e51733eca481a02bab731b7d030d74e400 Author: Dan Callaghan Date: Mon, 21 Apr 2025 13:56:59 +1000 update wiki submodule

fix signing key selection for sign_and_encrypt

2025-04-12T05:28:19Z

commit 104e89e51733eca481a02bab731b7d030d74e400 parent 07ae55696a4121295386b4cd6ae498b2c00254a0 Author: Dan Callaghan Date: Sat, 12 Apr 2025 15:28:19 +1000 fix signing key selection for sign_and_encrypt The GPGME::Crypto#encrypt() method looks for :signers in its options hash, but Sup was passing :signer which has no effect. As a result, the mail would always be signed with the user's default signing key. Keep the :signer misspelling on the Sup side, to avoid breaking anyone's existing 'gpg-options' hooks, but pass :signers to GPGME. Fixes #522.

contrib/nix: patch rmail for frozen string literals

2025-04-10T09:44:22Z

commit 07ae55696a4121295386b4cd6ae498b2c00254a0 parent e9f8f8439973152046f4a0509efd61b0dd4263c2 Author: Dan Callaghan Date: Thu, 10 Apr 2025 19:44:22 +1000 contrib/nix: patch rmail for frozen string literals

contrib/nix: also bump unicode gem for older Rubies

2025-04-06T05:05:27Z

commit e9f8f8439973152046f4a0509efd61b0dd4263c2 parent a0a215ba3b9bd67c3a32c5865b990c91637896a2 Author: Dan Callaghan Date: Sun, 6 Apr 2025 15:05:27 +1000 contrib/nix: also bump unicode gem for older Rubies Commit 58d53188 bumped unicode from 0.4.4.4 to 0.4.4.5, to fix a compilation failure with newer clang, but only in Gemfile.lock. Bump it in the older Rubies' pinned lockfiles as well for consistency.

depend on benchmark and fiddle gems

2025-04-06T04:40:18Z

commit a0a215ba3b9bd67c3a32c5865b990c91637896a2 parent de66aa76ac4321d1d4495dee2245a52d5f5af691 Author: Dan Callaghan Date: Sun, 6 Apr 2025 14:40:18 +1000 depend on benchmark and fiddle gems Fixes these warnings on Ruby 3.4: lib/sup/util.rb:10: warning: benchmark was loaded from the standard library, but will no longer be part of the default gems starting from Ruby 3.5.0. You can add benchmark to your Gemfile or gemspec to silence this warning. lib/sup/util/locale_fiddler.rb:5: warning: fiddle was loaded from the standard library, but will no longer be part of the default gems starting from Ruby 3.5.0. You can add fiddle to your Gemfile or gemspec to silence this warning.

contrib/nix: bump all gem dependencies to their latest versions

2025-04-06T01:44:05Z

commit de66aa76ac4321d1d4495dee2245a52d5f5af691 parent 7a444112c8d915515d6a90644cf2b4e86000fa30 Author: Dan Callaghan Date: Sun, 6 Apr 2025 11:44:05 +1000 contrib/nix: bump all gem dependencies to their latest versions ... except rubocop, because newer versions complain about this: https://docs.rubocop.org/rubocop/plugin_migration_guide.html which will require me to either skip it on older Rubies or upgrade all of them to the latest rubocop.

contrib/nix: use patched ncursesw-ruby instead of modifying ncurses

2025-04-06T00:20:20Z

commit 7a444112c8d915515d6a90644cf2b4e86000fa30 parent a9e44f2145a56f88286c98aeda62ab0ca2dacc87 Author: Dan Callaghan Date: Sun, 6 Apr 2025 10:20:20 +1000 contrib/nix: use patched ncursesw-ruby instead of modifying ncurses

contrib/nix: switch to official release of xapian-ruby 1.4.27

2025-04-05T23:50:26Z

commit a9e44f2145a56f88286c98aeda62ab0ca2dacc87 parent f7ade4e96d9a1a7198b0f84644f62ab58bd5536d Author: Dan Callaghan Date: Sun, 6 Apr 2025 09:50:26 +1000 contrib/nix: switch to official release of xapian-ruby 1.4.27

fix some frozen string literal warnings

2025-04-05T04:50:50Z

commit f7ade4e96d9a1a7198b0f84644f62ab58bd5536d parent 1a8e63f3fe2549ad27684192a6fd80bc0dbdce24 Author: Dan Callaghan Date: Sat, 5 Apr 2025 15:50:50 +1100 fix some frozen string literal warnings

contrib/nix: build ncurses with --disable-opaque-curses

2025-04-05T04:04:33Z

commit 1a8e63f3fe2549ad27684192a6fd80bc0dbdce24 parent 69a14cb07502651cd2997cf0578fa98e406ce222 Author: Dan Callaghan Date: Sat, 5 Apr 2025 15:04:33 +1100 contrib/nix: build ncurses with --disable-opaque-curses Works around #623.

contrib/nix: add some more MacOS platforms to Gemfile.lock

2025-03-30T02:23:14Z

commit 69a14cb07502651cd2997cf0578fa98e406ce222 parent 719ad2d9a6926f6482ef02385217a9aecec8a282 Author: Dan Callaghan Date: Sun, 30 Mar 2025 13:23:14 +1100 contrib/nix: add some more MacOS platforms to Gemfile.lock See if this will satisfy the Github Actions MacOS runners.

Github Actions: skip Ruby < 3.1 on MacOS

2025-03-30T02:11:50Z

commit 719ad2d9a6926f6482ef02385217a9aecec8a282 parent 0c04d2584402ad51451007496569b25890b6864a Author: Dan Callaghan Date: Sun, 30 Mar 2025 13:11:50 +1100 Github Actions: skip Ruby < 3.1 on MacOS It's too much effort to keep everything building.

xapian-ruby: 1.4.22 → 1.4.27

2025-03-30T02:11:43Z

commit 0c04d2584402ad51451007496569b25890b6864a parent 5b149ef78a9f2d9e325c804671d2595e06c0d3fd Author: Dan Callaghan Date: Sun, 30 Mar 2025 13:11:43 +1100 xapian-ruby: 1.4.22 → 1.4.27

contrib/nix: use pinned gemset for Ruby 2.7 and 3.0

2025-03-30T01:17:50Z

commit 5b149ef78a9f2d9e325c804671d2595e06c0d3fd parent 9933a555c07eae0ba5770e7300cdbccc58c85907 Author: Dan Callaghan Date: Sun, 30 Mar 2025 12:17:50 +1100 contrib/nix: use pinned gemset for Ruby 2.7 and 3.0 They were already pinned to an older nixpkgs commit, so their gemset needs to be pinned as well. I can't upgrade xapian-ruby on these old versions because the xapian in their nixpkgs is too old.

Github Actions: update to nix-quick-install-action v30

2025-03-29T08:29:13Z

commit 9933a555c07eae0ba5770e7300cdbccc58c85907 parent 8c4fa29f77f4782f88548926e0a329aac2bb3b51 Author: Dan Callaghan Date: Sat, 29 Mar 2025 19:29:13 +1100 Github Actions: update to nix-quick-install-action v30 Trying to solve this failure on Ubuntu runners: installing 'nix-2.12.0' error: cannot set loopback interface flags: Operation not permitted (use '--show-trace' to show detailed location information) Error: Process completed with exit code 1.

test against Ruby 3.4 in CI

2025-03-29T07:31:18Z

commit 8c4fa29f77f4782f88548926e0a329aac2bb3b51 parent 07e633cb6af7f2cb947143b64a29e774b61d3c20 Author: Dan Callaghan Date: Sat, 29 Mar 2025 18:31:18 +1100 test against Ruby 3.4 in CI Also add Ruby 3.3 to my runner script. Evidently I forgot this back when I added it to Github Actions.

contrib/nix: bump nixpkgs

2025-03-29T07:19:05Z

commit 07e633cb6af7f2cb947143b64a29e774b61d3c20 parent 58d531881595bf99626f01e8c0db2e71c3306858 Author: Dan Callaghan Date: Sat, 29 Mar 2025 18:19:05 +1100 contrib/nix: bump nixpkgs This is to get ruby_3_4.

unicode: 0.4.4.4 → 0.4.4.5

2025-03-29T07:02:39Z

commit 58d531881595bf99626f01e8c0db2e71c3306858 parent 26f5b3361a015e6be63aa372261a0da3af3be037 Author: Dan Callaghan Date: Sat, 29 Mar 2025 18:02:39 +1100 unicode: 0.4.4.4 → 0.4.4.5

fix tests for GPG signature verification

2024-10-14T07:43:09Z

commit 26f5b3361a015e6be63aa372261a0da3af3be037 parent b22a9fb3dd2d31ce805ea51581de5a51ba413839 Author: Dan Callaghan Date: Mon, 14 Oct 2024 18:43:09 +1100 fix tests for GPG signature verification They were not checking that the verification actually succeeded, nor which key was being used. Regenerate test keys with a new ECC key for sup-fake-ecc@fake.fake, which is used by the test_verify_unknown_keytype test case.

tests: skip instead of passing if GPG is not working

2024-07-14T07:29:46Z

commit b22a9fb3dd2d31ce805ea51581de5a51ba413839 parent ac12be507b1473e12004f552bf34e8b4d8209532 Author: Dan Callaghan Date: Sun, 14 Jul 2024 17:29:46 +1000 tests: skip instead of passing if GPG is not working

encode attachments if they have lines longer than 998 chars

2024-07-07T12:21:57Z

commit ac12be507b1473e12004f552bf34e8b4d8209532 parent f3b63120aa7c7610cf691f31e5998988ff813586 Author: Dan Callaghan Date: Sun, 7 Jul 2024 22:21:57 +1000 encode attachments if they have lines longer than 998 chars As per: https://datatracker.ietf.org/doc/html/rfc5322#section-2.1.1 Fixes #489.

add a test case for 8bit attachment without GPG

2024-07-07T08:14:05Z

commit f3b63120aa7c7610cf691f31e5998988ff813586 parent f16d47380bf36abac12ae2994ba5b3a252dcf947 Author: Dan Callaghan Date: Sun, 7 Jul 2024 18:14:05 +1000 add a test case for 8bit attachment without GPG This is the opposite behaviour to the test case added in the previous commit (the attachment goes in unmodified).

use MacOS 13 in Github Actions

2024-07-07T07:09:53Z

commit f16d47380bf36abac12ae2994ba5b3a252dcf947 parent d02f27488918e848112e527499f83c66130c0782 Author: Dan Callaghan Date: Sun, 7 Jul 2024 17:09:53 +1000 use MacOS 13 in Github Actions MacOS 14 is failing due to: https://github.com/nixbuild/nix-quick-install-action/issues/43

don't add multiple contradictory Content-Transfer-Encoding headers

2024-07-07T06:48:05Z

commit d02f27488918e848112e527499f83c66130c0782 parent e2bdf6ac790fc2926ee0d8da2685ba2cadca253b Author: Dan Callaghan Date: Sun, 7 Jul 2024 16:48:05 +1000 don't add multiple contradictory Content-Transfer-Encoding headers Fixes #502.

treat message/global as an embedded message per RFC6532

2024-05-12T07:52:15Z

commit e2bdf6ac790fc2926ee0d8da2685ba2cadca253b parent 7085ec55ef5bd7d770fead5c6e16000ebc1a2629 Author: Dan Callaghan Date: Sun, 12 May 2024 17:52:15 +1000 treat message/global as an embedded message per RFC6532 Fixes #602.

use a fallback date if Date header is not valid RFC2822

2024-05-12T06:21:47Z

commit 7085ec55ef5bd7d770fead5c6e16000ebc1a2629 parent 7a035e6a53449b77cceae1d8045c51c4063b0ab6 Author: Dan Callaghan Date: Sun, 12 May 2024 16:21:47 +1000 use a fallback date if Date header is not valid RFC2822 Use Time.rfc2822 to parse the Date header in RFC2822 format, rather than letting Time.parse try heuristics to guess the format. If parsing fails, use the fallback date (mtime or From line timestamp) instead of letting Time.parse fill in the missing pieces from the current time. Fixes #252.

use a fallback date if Date header is missing

2024-05-12T04:07:11Z

commit 7a035e6a53449b77cceae1d8045c51c4063b0ab6 parent 4255437a5b252c7268c7ec2743b182982334ccb5 Author: Dan Callaghan Date: Sun, 12 May 2024 14:07:11 +1000 use a fallback date if Date header is missing If a message doesn't have a usable Date header, we can fall back to using the modtime of the file for Maildir, or the delivery time which should have been written to the From line in mbox. If all else fails, use a fixed date in the past, rather than Time.now. This will prevent the mails from popping back to the top of the thread list every time Sup reloads them. Fixes #596.

use Time.strptime when parsing mbox From lines

2024-05-12T04:02:57Z

commit 4255437a5b252c7268c7ec2743b182982334ccb5 parent 6227c9ed977e0c88c9a499cccca4c1383ee2f3bb Author: Dan Callaghan Date: Sun, 12 May 2024 14:02:57 +1000 use Time.strptime when parsing mbox From lines

contrib/nix: add back platforms to Gemfile.lock

2024-04-21T02:25:30Z

commit 6227c9ed977e0c88c9a499cccca4c1383ee2f3bb parent 104f56f57e9131ea029f2cdf7d3f648e1156e492 Author: Dan Callaghan Date: Sun, 21 Apr 2024 12:25:30 +1000 contrib/nix: add back platforms to Gemfile.lock Accidentally lost them in commit d6bcac5a. We need these to satisfy Github Actions.

prepare for next release

2024-04-21T01:56:26Z

commit 104f56f57e9131ea029f2cdf7d3f648e1156e492 parent 76d6195181cec1b1b4d7b7316f8b20ef238df95f Author: Dan Callaghan Date: Sun, 21 Apr 2024 11:56:26 +1000 prepare for next release

release 1.2

2024-04-21T01:55:30Z

commit 76d6195181cec1b1b4d7b7316f8b20ef238df95f parent d6bcac5aab6dad4a42e86a258ce15447adf0c791 Author: Dan Callaghan Date: Sun, 21 Apr 2024 11:55:30 +1000 release 1.2

bump all dependencies to their latest versions

2024-04-21T01:13:51Z

commit d6bcac5aab6dad4a42e86a258ce15447adf0c791 parent 57ce05a58b644352588f46274fcb3a55dbe4b234 Author: Dan Callaghan Date: Sun, 21 Apr 2024 11:13:51 +1000 bump all dependencies to their latest versions

ncursesw: 1.4.10 → 1.4.11

2024-04-21T00:06:28Z

commit 57ce05a58b644352588f46274fcb3a55dbe4b234 parent 3c63d7c6d66fa99d6bdecf6484cdb6df6d8c4621 Author: Dan Callaghan Date: Sun, 21 Apr 2024 10:06:28 +1000 ncursesw: 1.4.10 → 1.4.11

contrib/nix: work around clang errors in unicode gem

2024-04-14T04:32:48Z

commit 3c63d7c6d66fa99d6bdecf6484cdb6df6d8c4621 parent 6f91b0610063bcb47a64d37ffab5d3b2bb491f68 Author: Dan Callaghan Date: Sun, 14 Apr 2024 14:32:48 +1000 contrib/nix: work around clang errors in unicode gem Same new clang error as in ncursesw, but in this case it's just missing casts, not misdeclared arguments. There's no clear upstream to publish a proper fix so let's just suppress the error for now so that the gem build on MacOS can succeed.

contrib/nix: patch ncursesw gem to fix method args

2024-04-14T03:46:47Z

commit 6f91b0610063bcb47a64d37ffab5d3b2bb491f68 parent 2d5a6951fed09afbc8d2e27a64b85a31eb5e648d Author: Dan Callaghan Date: Sun, 14 Apr 2024 13:46:47 +1000 contrib/nix: patch ncursesw gem to fix method args Another temporary patch until I release ncursesw. Should fix the build on MacOS, where clang 16 is complaining about mismatched function pointer types.

fix manifest

2024-04-13T08:11:06Z

commit 2d5a6951fed09afbc8d2e27a64b85a31eb5e648d parent 36464da82e198a0272efebda209eb610fdfa6d12 Author: Dan Callaghan Date: Sat, 13 Apr 2024 18:11:06 +1000 fix manifest Sigh, when will I learn...

contrib/nix: add a shell.nix for testing 'gem install'

2024-04-13T07:57:37Z

commit 36464da82e198a0272efebda209eb610fdfa6d12 parent 492091ad4d544da7fce650b70347a151f5edb053 Author: Dan Callaghan Date: Sat, 13 Apr 2024 17:57:37 +1000 contrib/nix: add a shell.nix for testing 'gem install'

conditionally install xapian-ruby gem based on an environment variable

2024-04-13T07:37:21Z

commit 492091ad4d544da7fce650b70347a151f5edb053 parent d996908a2c56e305bd21d18d5dcd16a5a0a93c02 Author: Dan Callaghan Date: Sat, 13 Apr 2024 17:37:21 +1000 conditionally install xapian-ruby gem based on an environment variable ... instead of checking if the host is OpenBSD. There are other situations where the xapian-ruby gem may be broken and the user may prefer to install it themselves, such as on recent MacOS.

update wiki submodule

2024-04-07T07:21:48Z

commit d996908a2c56e305bd21d18d5dcd16a5a0a93c02 parent 40da2bbec0a5fd2c09d46ef7c7d1fd46d83ca38d Author: Dan Callaghan Date: Sun, 7 Apr 2024 17:21:48 +1000 update wiki submodule

contrib/nix: patch ncursesw gem to undefine allocator

2024-04-07T05:56:13Z

commit 40da2bbec0a5fd2c09d46ef7c7d1fd46d83ca38d parent 1953c3f3a0a049a8605d103d493e19b263b6e07a Author: Dan Callaghan Date: Sun, 7 Apr 2024 15:56:13 +1000 contrib/nix: patch ncursesw gem to undefine allocator Fixes "undefining the allocator of T_DATA class" warnings from ncursesw. This patch is temporary until I release a new version of the ncursesw gem.

remove link to IRC channel

2024-04-07T04:32:27Z

commit 1953c3f3a0a049a8605d103d493e19b263b6e07a parent a5aa56f037e77b5b4bd82434a3bc2e25446c758a Author: Dan Callaghan Date: Sun, 7 Apr 2024 14:32:27 +1000 remove link to IRC channel Nobody's in the IRC channel anymore I think (I'm certainly not).

fix rescue path for OutOfSyncSourceError

2024-04-07T01:20:03Z

commit a5aa56f037e77b5b4bd82434a3bc2e25446c758a parent bddd7f096cd836fa18221116dcc99aa3257c91ab Author: Dan Callaghan Date: Sun, 7 Apr 2024 11:20:03 +1000 fix rescue path for OutOfSyncSourceError This rescue block was supposed to let sup recover from a failure to load a message, and just show an error in place of the body. But in the case of OutOfSyncSourceError (which might be the only error we hit here?) the rescue path itself triggered another OutOfSyncSourceError because of its log messages. --- Redwood::OutOfSyncSourceError from thread: load messages for thread-view-mode Redwood::OutOfSyncSourceError /home/dan/sup/lib/sup/message.rb:246:in `location' /home/dan/sup/lib/sup/message.rb:763:in `warn_with_location' /home/dan/sup/lib/sup/message.rb:273:in `rescue in load_from_source!' /home/dan/sup/lib/sup/message.rb:260:in `load_from_source!' /home/dan/sup/lib/sup/modes/thread_index_mode.rb:117:in `block (3 levels) in select' /home/dan/sup/lib/sup/thread.rb:70:in `block in each' /home/dan/sup/lib/sup/thread.rb:176:in `each_with_stuff' /home/dan/sup/lib/sup/thread.rb:69:in `each' /home/dan/sup/lib/sup/modes/thread_index_mode.rb:114:in `each_with_index' /home/dan/sup/lib/sup/modes/thread_index_mode.rb:114:in `block (2 levels) in select' /home/dan/sup/lib/sup/buffer.rb:700:in `say' (eval at /home/dan/sup/lib/sup/util.rb:580):1:in `say' /home/dan/sup/lib/sup/modes/thread_index_mode.rb:113:in `block in select' /home/dan/sup/lib/sup.rb:84:in `block in reporting_thread' Fix the logging to avoid triggering more OutOfSyncSourceErrors.

also test on Ruby 3.3

2024-01-01T07:49:47Z

commit bddd7f096cd836fa18221116dcc99aa3257c91ab parent d3288571a5d4d7a44eb5b4e4393209ba71998114 Author: Dan Callaghan Date: Mon, 1 Jan 2024 18:49:47 +1100 also test on Ruby 3.3

contrib/nix: bump nixpkgs

2024-01-01T06:12:21Z

commit d3288571a5d4d7a44eb5b4e4393209ba71998114 parent 156ba2416e4b660f09846d21f0d24b59c565154a Author: Dan Callaghan Date: Mon, 1 Jan 2024 17:12:21 +1100 contrib/nix: bump nixpkgs This is to get ruby_3_3.

rubocop: 1.32.0 → 1.59.0

2024-01-01T10:52:38Z

commit 156ba2416e4b660f09846d21f0d24b59c565154a parent e7f1796251de252d54984473d1722a9995dbce80 Author: Dan Callaghan Date: Mon, 1 Jan 2024 21:52:38 +1100 rubocop: 1.32.0 → 1.59.0 To get this Ruby 3.3 fix: https://github.com/whitequark/parser/pull/929

xapian-ruby: 1.4.18 → 1.4.22

2024-01-01T05:56:51Z

commit e7f1796251de252d54984473d1722a9995dbce80 parent f643412f0f466d93ec0ce00b1283824951989600 Author: Dan Callaghan Date: Mon, 1 Jan 2024 16:56:51 +1100 xapian-ruby: 1.4.18 → 1.4.22 This fixes some "undefining the allocator of T_DATA class" warnings. Leave Ruby 2.6 using 1.4.18 because 1.4.22 does not compile against the older Xapian in that old revision of nixpkgs.

gpgme: 2.0.20 → 2.0.23

2024-01-01T05:16:31Z

commit f643412f0f466d93ec0ce00b1283824951989600 parent c7e5fdde22fbb0371d0123d8efc76238e69372c8 Author: Dan Callaghan Date: Mon, 1 Jan 2024 16:16:31 +1100 gpgme: 2.0.20 → 2.0.23 This fixes some "undefining the allocator of T_DATA class" warnings: https://github.com/ueno/ruby-gpgme/commit/ac92f1428ed692e08a297ba3d0624d1edecb6245

contrib/nix: add a helper script to run checks across all Ruby versions

2024-01-01T03:48:57Z

commit c7e5fdde22fbb0371d0123d8efc76238e69372c8 parent 0d77240ad21004ec99655223f230682b28af16dd Author: Dan Callaghan Date: Mon, 1 Jan 2024 14:48:57 +1100 contrib/nix: add a helper script to run checks across all Ruby versions

contrib/nix: add a note to remind myself how to invoke bundix

2024-01-01T02:54:33Z

commit 0d77240ad21004ec99655223f230682b28af16dd parent a6a3270f39f4e7c6c45ab9e8a352d1d2591c0699 Author: Dan Callaghan Date: Mon, 1 Jan 2024 13:54:33 +1100 contrib/nix: add a note to remind myself how to invoke bundix

fix unused variable warnings in Thread

2024-01-01T02:47:17Z

commit a6a3270f39f4e7c6c45ab9e8a352d1d2591c0699 parent 88779f92368cb85057d10dce751c04ef15cee803 Author: Dan Callaghan Date: Mon, 1 Jan 2024 13:47:17 +1100 fix unused variable warnings in Thread

also test against Ruby 3.2 in CI

2023-03-29T10:41:13Z

commit 88779f92368cb85057d10dce751c04ef15cee803 parent f1f44a18ea35575c95767b96a37cfaf47bf7d95a Author: Dan Callaghan Date: Wed, 29 Mar 2023 21:41:13 +1100 also test against Ruby 3.2 in CI

contrib/nix: bump nixpkgs

2023-03-29T08:50:12Z

commit f1f44a18ea35575c95767b96a37cfaf47bf7d95a parent d2a2a0d7544237a29bf932773cc85e9050bd9c26 Author: Dan Callaghan Date: Wed, 29 Mar 2023 19:50:12 +1100 contrib/nix: bump nixpkgs This is to get ruby_3_2.

treat empty URI#host the same as nil

2023-03-29T10:37:16Z

commit d2a2a0d7544237a29bf932773cc85e9050bd9c26 parent 4e7b5c9a0fedeece3adeaecf79ec2088082c5fbc Author: Dan Callaghan Date: Wed, 29 Mar 2023 21:37:16 +1100 treat empty URI#host the same as nil Since Ruby 3.2, the URI#host method returns empty string rather than nil if the URI does not have a host component (like maildir:///blah). Apparently for some kind of security reasons, see: https://github.com/ruby/ruby/commit/dd5118f8524c425894d4716b787837ad7380bb0d

load sources.yaml relative to index dir

2023-03-29T10:16:11Z

commit 4e7b5c9a0fedeece3adeaecf79ec2088082c5fbc parent b21f0328661f17285fd3ba321a0ad0d7dc405337 Author: Dan Callaghan Date: Wed, 29 Mar 2023 21:16:11 +1100 load sources.yaml relative to index dir This only matters in the test suite, which is the only time that Index is constructed pointing to a custom directory (the test's temp dir). Fixes test/integration/test_mbox.rb and test/integration/test_maildir.rb to not try reading my sources from ~/.sup during test execution.

ensure all tests can be run individually

2023-03-29T10:06:12Z

commit b21f0328661f17285fd3ba321a0ad0d7dc405337 parent a20a7fec266fbf4ee89876231c49b7027b69c136 Author: Dan Callaghan Date: Wed, 29 Mar 2023 21:06:12 +1100 ensure all tests can be run individually The Redwood module was leaking into other tests via the wrongly placed `include` in test/test_header_parsing.rb. Without this fix, running `rake test TEST=test/integration/test_maildir.rb` or `rake test TEST=test/integration/test_mbox.rb` would fail because Redwood was not being included in that case. Fix the tests to include Redwood into each individual class where it's needed, and avoid leaking into other tests.

cache nix store in CI

2023-03-05T10:39:11Z

commit a20a7fec266fbf4ee89876231c49b7027b69c136 parent 60fa34485999583aedd5f8625883d184ff98c38d Author: Dan Callaghan Date: Sun, 5 Mar 2023 21:39:11 +1100 cache nix store in CI

use nix-shell in CI

2023-03-04T09:24:45Z

commit 60fa34485999583aedd5f8625883d184ff98c38d parent 21701eae2257e78bdc099dd384d516034fee243c Author: Dan Callaghan Date: Sat, 4 Mar 2023 20:24:45 +1100 use nix-shell in CI

add nix-shell expressions

2023-03-05T04:36:42Z

commit 21701eae2257e78bdc099dd384d516034fee243c parent 00d2dc2da4a026a809250100697d204a633f9fdb Author: Dan Callaghan Date: Sun, 5 Mar 2023 15:36:42 +1100 add nix-shell expressions One for each version of Ruby back to Ruby 2.4, so that I can run sup and its tests under all these versions reproducibly. These should also work on MacOS, so they can be used for CI and they will give me a greater chance of reproducing and fixing the problems there, rather than using all that Homebrew stuff that I am clueless about. The separate copy of the Gemfile is needed because bundix cannot understand 'gemspec' calls in Gemfiles, it needs the dependencies to be spelled out explicitly. So this Gemfile is just a copy-paste of the same dependency declarations from sup.gemspec.

stop testing Ruby < 2.4 in Github Actions

2023-03-03T04:06:51Z

commit 00d2dc2da4a026a809250100697d204a633f9fdb parent f6142cc42d8b85ba053f4866f1887de33a693022 Author: Dan Callaghan Date: Fri, 3 Mar 2023 15:06:51 +1100 stop testing Ruby < 2.4 in Github Actions It's increasingly difficult to get a reproducible, working environment for these old Ruby versions. Some gems (highline, unicode-display-width) have begun asserting their Ruby interpreter version at runtime. And some things (gpgme) fail to compile in newer environments with hard to fix errors. Even using old revisions of nixpkgs it's not possible because bundler < 2.2 was broken for a long time in nixpkgs.

replace 2&>1 by 2>&1

2023-03-01T19:22:09Z

commit f6142cc42d8b85ba053f4866f1887de33a693022 parent caa5c663befb123d158aa928716c194a11e8f51d Author: Patrice Duroux Date: Wed, 1 Mar 2023 20:22:09 +0100 replace 2&>1 by 2>&1 Hi, Just in case. ps: it follows the point here: https://lists.debian.org/debian-qa/2023/02/msg00052.html

scrub strings in fix_encoding! rather than triple-encoding

2022-05-23T10:20:46Z

commit caa5c663befb123d158aa928716c194a11e8f51d parent acae15af917fd4a27a924dcfe31d52bfe48bee22 Author: Dan Callaghan Date: Mon, 23 May 2022 20:20:46 +1000 scrub strings in fix_encoding! rather than triple-encoding This will have the same effect, but it should be more efficient since in the normal case there will be no extra allocations. This matters for fix_encoding! since it's applied to a lot of strings everywhere.

console mode: fix deprecated usage of Enumerator.new

2022-06-18T07:44:36Z

commit acae15af917fd4a27a924dcfe31d52bfe48bee22 parent 3a7fdaed0c236188be4e85e90782e7b1962016ee Author: Dan Callaghan Date: Sat, 18 Jun 2022 17:44:36 +1000 console mode: fix deprecated usage of Enumerator.new The 'query' method suffered a similar issue as #603, with a similar fix as in PR #608.

tests: add coverage for sup-sync-back-maildir

2022-06-18T08:00:13Z

commit 3a7fdaed0c236188be4e85e90782e7b1962016ee parent ffd3959caf7226d2ac188f83db819c63e753c3b6 Author: Dan Callaghan Date: Sat, 18 Jun 2022 18:00:13 +1000 tests: add coverage for sup-sync-back-maildir

Fix usage of API deprecated in Ruby 3

2022-06-11T17:08:29Z

commit ffd3959caf7226d2ac188f83db819c63e753c3b6 parent 7b501064f4e42c478b6e34f567f0f4b96b49e16a Author: gclv Date: Sat, 11 Jun 2022 18:08:29 +0100 Fix usage of API deprecated in Ruby 3

History.txt: add a few missing attributions

2022-05-23T11:04:24Z

commit 7b501064f4e42c478b6e34f567f0f4b96b49e16a parent 29253088d5f4bd778f2268792cf4b37a716d84b7 Author: Dan Callaghan Date: Mon, 23 May 2022 21:04:24 +1000 History.txt: add a few missing attributions Starting to feel a bit redundant since I'm the only one sending PRs anymore, but oh well...

prepare for next release

2022-05-23T10:56:53Z

commit 29253088d5f4bd778f2268792cf4b37a716d84b7 parent 4ab9f262fad6ee2660ce4b5b638827076ed673c4 Author: Dan Callaghan Date: Mon, 23 May 2022 20:56:53 +1000 prepare for next release

release 1.1

2022-05-23T10:49:53Z

commit 4ab9f262fad6ee2660ce4b5b638827076ed673c4 parent 69be57859d18b7f0be78c0e6627b6f47541261b6 Author: Dan Callaghan Date: Mon, 23 May 2022 20:49:53 +1000 release 1.1

send :prepend instead of calling it

2022-05-23T10:09:27Z

commit 69be57859d18b7f0be78c0e6627b6f47541261b6 parent 7bcb85006d3b4f58cb1566331f378298b8c95fb4 Author: Dan Callaghan Date: Mon, 23 May 2022 20:09:27 +1000 send :prepend instead of calling it For compatibility with Ruby 2.0, where 'prepend' is private.

tests: delete a dead integration test

2022-05-23T09:13:04Z

commit 7bcb85006d3b4f58cb1566331f378298b8c95fb4 parent 5606b05134d1edc997bb6d0895138d8782edd8cf Author: Dan Callaghan Date: Mon, 23 May 2022 19:13:04 +1000 tests: delete a dead integration test

tests: add some missing assertions

2022-05-23T09:08:40Z

commit 5606b05134d1edc997bb6d0895138d8782edd8cf parent 7902898566c3a28e9391ee4f9f9adb3ffcce51e1 Author: Dan Callaghan Date: Mon, 23 May 2022 19:08:40 +1000 tests: add some missing assertions Doesn't seem like a particularly interesting test case, and the git history doesn't give any clues about what it's trying to cover. But let's just fill in some assertions for completeness' sake.

tests: capture subprocess output from sup-add

2022-05-23T09:03:03Z

commit 7902898566c3a28e9391ee4f9f9adb3ffcce51e1 parent 5fdcfa1bd6520c72b7a53d2b425a221f9f8ff268 Author: Dan Callaghan Date: Mon, 23 May 2022 19:03:03 +1000 tests: capture subprocess output from sup-add

tests: capture expected warning messages

2022-05-09T11:23:06Z

commit 5fdcfa1bd6520c72b7a53d2b425a221f9f8ff268 parent 4a77a4741ea907b193b04fd52da2715ebfd0ceb8 Author: Dan Callaghan Date: Mon, 9 May 2022 21:23:06 +1000 tests: capture expected warning messages

Socket.gethostbyname -> Addrinfo.getaddrinfo

2022-05-09T10:42:11Z

commit 4a77a4741ea907b193b04fd52da2715ebfd0ceb8 parent 625148ecb7e37334619838dd274f43e73635d9d7 Author: Dan Callaghan Date: Mon, 9 May 2022 20:42:11 +1000 Socket.gethostbyname -> Addrinfo.getaddrinfo This seems like an unreliable way to guess the host's FQDN but I suppose it's better than nothing...

fix "warning: method redefined"

2022-05-09T10:29:46Z

commit 625148ecb7e37334619838dd274f43e73635d9d7 parent 11ed9754e0d6e20c81ec522138cd592fd963c94f Author: Dan Callaghan Date: Mon, 9 May 2022 20:29:46 +1000 fix "warning: method redefined"

restore monkey-patching of RMail::Serialize.calculate_boundaries

2022-05-09T10:22:06Z

commit 11ed9754e0d6e20c81ec522138cd592fd963c94f parent a8dfda2ebf2404ed93a272cff374fed76baa53db Author: Dan Callaghan Date: Mon, 9 May 2022 20:22:06 +1000 restore monkey-patching of RMail::Serialize.calculate_boundaries The monkey-patch was not applying due to module load ordering. It seems to still be needed in order to avoid superfluous MIME-Version headers on nested signed or encrypted parts, although such headers are likely not harmful. Includes test cases which attempt to reproduce the original behaviour and fix that commit 07f0be01 probably intended.

remove unused RMail::Header monkey-patch

2022-05-08T01:39:10Z

commit a8dfda2ebf2404ed93a272cff374fed76baa53db parent 3a97fd68ae769e68f2e0c7cf4caf7e0564d069b2 Author: Dan Callaghan Date: Sun, 8 May 2022 11:39:10 +1000 remove unused RMail::Header monkey-patch This was not actually applying, because of module load order, and it didn't actually work anyway ('value' is not defined). It's not needed because RMail header values are always stripped, they can never consist of only whitespace. Thus the existing 'content_type' method in RMail works fine.

relax minitest requirement

2022-05-07T23:55:39Z

commit 3a97fd68ae769e68f2e0c7cf4caf7e0564d069b2 parent 9a69f7b6dba266ed6624a758e4e12a6c286857f7 Author: Dan Callaghan Date: Sun, 8 May 2022 09:55:39 +1000 relax minitest requirement Minitest 5.15.0 works well.

fix "warning: ambiguity between regexp and two divisions"

2022-05-07T23:52:11Z

commit 9a69f7b6dba266ed6624a758e4e12a6c286857f7 parent 68fcb8ec7763739ddcdf570a1df301129695a3ca Author: Dan Callaghan Date: Sun, 8 May 2022 09:52:11 +1000 fix "warning: ambiguity between regexp and two divisions"

fix "warning: `*' interpreted as argument prefix"

2022-05-07T23:47:14Z

commit 68fcb8ec7763739ddcdf570a1df301129695a3ca parent c16623b75ad17a7689c6013d7e64d4a1108b8f5c Author: Dan Callaghan Date: Sun, 8 May 2022 09:47:14 +1000 fix "warning: `*' interpreted as argument prefix"

fix "warning: `&' interpreted as argument prefix"

2022-05-07T23:42:52Z

commit c16623b75ad17a7689c6013d7e64d4a1108b8f5c parent ab504e72382d8dc43ca67351b5eec7003fce994c Author: Dan Callaghan Date: Sun, 8 May 2022 09:42:52 +1000 fix "warning: `&' interpreted as argument prefix"

fix "warning: assigned but unused variable"

2022-05-07T23:07:54Z

commit ab504e72382d8dc43ca67351b5eec7003fce994c parent 8441116f67308574aef2339412b426afb151a399 Author: Dan Callaghan Date: Sun, 8 May 2022 09:07:54 +1000 fix "warning: assigned but unused variable"

update wiki submodule

2022-05-06T00:52:50Z

commit 8441116f67308574aef2339412b426afb151a399 parent d84d6f266e37cae8113099a5297e898585f41643 Author: Dan Callaghan Date: Fri, 6 May 2022 10:52:50 +1000 update wiki submodule

handle malformed UTF-7 in RFC2047 words

2022-05-05T01:09:05Z

commit d84d6f266e37cae8113099a5297e898585f41643 parent 283a8fe5a5f6c362452eb89e9e6f007ec51e822c Author: Dan Callaghan Date: Thu, 5 May 2022 11:09:05 +1000 handle malformed UTF-7 in RFC2047 words Turns out spammers are not very good at following standards. The code for handling UTF-7 added in commit 9b4d42a2 would crash like this: --- Encoding::UndefinedConversionError from thread: poll after loading inbox "\xE2" from ASCII-8BIT to UTF-8 /home/dan/sup/lib/sup/rfc2047.rb:58:in `encode' /home/dan/sup/lib/sup/rfc2047.rb:58:in `block in decode_to' /home/dan/sup/lib/sup/rfc2047.rb:31:in `gsub' /home/dan/sup/lib/sup/rfc2047.rb:31:in `decode_to' /home/dan/sup/lib/sup/message.rb:77:in `decode_header_field' /home/dan/sup/lib/sup/message.rb:81:in `block in parse_header' /home/dan/sup/lib/sup/util.rb:648:in `[]' /home/dan/sup/lib/sup/message.rb:117:in `parse_header' /home/dan/sup/lib/sup/message.rb:270:in `load_from_source!' [...] when given an RFC2047-encoded word which claims to be UTF-7 but contains high bytes, as in a spam subject line I recently received.