venta | Reading

Reading. Finished The Rose Field (Pullman)!!! I am Making Arrangements for it to Leave My House. ( Read more... )

ANYWAY. I finished it. It Is Done.

Then read the first few pages of Dead Hand Rule (Gladstone; latest in the Craft Wars) before deciding that actually I need to reread at least the end of Wicked Problems in order to remember what's going on...

Writing. Progress continues both glacial and extant.

Listening. My relisten-while-actually-awake of the first chunk of The Hidden Almanac continues, slowly.

Playing. We have finished an Exploders run on Hard in Inkulinati. I am contemplating, given how smoothly that went, whether I want to have a try at Very Hard...

Cooking. It's not quite "this week's breakfast dal, and a loaf of bread", but it does sort of feel like it was. Partly because for reasons we did not get our usual box of veg on Monday last week, which meant that we were scrabbling around using up Shelf Things and the occasional Supermarket Discount Item...

NO WAIT, I also DID make buckwheat pancakes, and inspired by lnr combined Tinned Pear and Stem Ginger with Vanilla Essence and also Ground Cardamom to go in same. V good. Will repeat.

Eating. My mother acquired for us, as A Special Treat, a variety of Baked Goods from The Fancy Bakery In Eddington: my favourite is still the fig-and-?ricotta, but the blueberry-and-?ricotta is also very good, as is the fougasse. A was extremely pleased with the pain aux raisins. AND my mother made some excellent baba ganoush, eaten with said fougasse.

This week also feat. rainbow bagels (which we got to watch some of the manufacturing process for!) as well as misc other foodstuffs from Shalom Hot Beigels.

A has some coffee and butterscotch cake (leftovers from a test bake!) from Flour Arrangements; alas by the time I got my act together to actually collect Excess Test Cake the apple pie and lemon had both all gone...

Exploring. I got to spend a little time in the City of London Cemetery, which is currently ablaze with (among other things) purple crocuses; we also (on our second attempt) managed to go on A Snowdrop Walk Around Anglesey (with thanks to aldabra for reminding me that it is That Time Of Year still!). Snowdrops excellent. May or may not get around to sharing some photos. (Our first attempt at A Snowdrop Walk Around Anglesey Abbey wound up mutating into a poke around the back of Churchill and Astronomy to peer at bulbs and other plants misc, which was also very enjoyable even if I did once again fail to take A to see the Barbara Hepworth.)

Growing. ... I bought a bag of snowdrops In The Green at Anglesey, to go into the ground around the cherry tree at the allotment? The lemongrass seedlings haven't all died?

diffrentcolours

For the last couple of weeks I've been telling every canvasser that's come to my door the same thing - that I'd like to vote Green, but I'm going to vote to keep Reform out, and that there wasn't any clear evidence of the best way to do that. Labour have been relying on an opinion poll of 150 people taken weeks ago before the close of nominations, which shows them in second place to Reform. The Greens have been relying on betting shop odds (very easy to manipulate) and internal polling (very unreliable). Both have been telling me that their "internal data" points to them being the best option.

Yesterday, a new poll from Omnisis was published. The sample size is still small (450) but much larger than we've seen before, and the polling was carried out after the candidates were finalised and after the campaign started. I am pleased to say that it puts Greens in front of Reform in the constituency, within margin of error. Labour are trailing Reform, also within margin of error, but the Green-Labour lead is outside that margin. It also says that the Green supporters are more likely to actually turn out and vote than Labour ones, which suggests that the Green lead will be bigger on Thursday than this poll suggests.

I can nitpick some things about the poll such as the sample size, but this is the best indicator I'm likely to get before going to the polls on Thursday. And I'm glad that I have some reasonably solid evidence that convinces my head that I can vote with my heart, rather than just looking for something that reinforces what I want to be true. I know from previous experience that the Labour vote in Gorton is soft as anything, and it's nice to have evidence that the Green squeeze is actually working.

There are some interesting nuggets in the data which show how people either don't understand questions or give nonsense answers, like the Labour supporter who would vote Reform if they "knew" this was a Labour-Reform battle. There are a small minority of both Green and Labour supporters who would vote Reform to stop the other from winning. The Reform vote is largely unsqueezable - even if they think Reform can't win, most supporters would either vote for them, or not vote at all, than vote for either Greens or Labour. More would break Green though, which suggests the "vote to get Kier Starmer out" message is working.

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: 0e57a69f14581a09c64220452225efd10842985d https://github.com/dreamwidth/dreamwidth/commit/0e57a69f14581a09c64220452225efd10842985d Author: Mark Smith mark@dreamwidth.org Date: 2026-02-21 (Sat, 21 Feb 2026)

Changed paths: M cgi-bin/Plack/Middleware/DW/SubdomainFunction.pm

Log Message:

Tidy SubdomainFunction.pm

Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: 558458c9ab5c5548f20b6b380857292cc8350203 https://github.com/dreamwidth/dreamwidth/commit/558458c9ab5c5548f20b6b380857292cc8350203 Author: Mark Smith mark@dreamwidth.org Date: 2026-02-20 (Fri, 20 Feb 2026)

Changed paths: M cgi-bin/Plack/Middleware/DW/Auth.pm M cgi-bin/Plack/Middleware/DW/SubdomainFunction.pm

Log Message:

Handle userpics subdomain in Plack middleware

The 'userpics' SUBDOMAIN_FUNCTION (e.g., v.dreamwidth.org) was not handled by SubdomainFunction middleware, causing userpic URLs to fall through to the Auth middleware which triggered unnecessary domain session cookie bounces. Under Apache, userpic_trans handles these directly without any auth flow.

Rewrite PATH_INFO from /{picid}/{userid} to /userpic/{picid}/{userid} to match DW::Controller::Userpic's route, and skip the domain session bounce since userpics are public images that don't need authentication.

Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: 0c0f75384cb09c6515b5381d9873105581cad595 https://github.com/dreamwidth/dreamwidth/commit/0c0f75384cb09c6515b5381d9873105581cad595 Author: Mark Smith mark@dreamwidth.org Date: 2026-02-20 (Fri, 20 Feb 2026)

Changed paths: M cgi-bin/LJ/Protocol.pm M cgi-bin/LJ/S2/ReplyPage.pm M cgi-bin/LJ/Talk.pm

Log Message:

Skip forced captcha on high-comment entries less than 30 days old

The 5k-comment captcha threshold was meant to slow bots targeting specific entries, but most spam hits old/abandoned journals while active anon memes are the ones suffering from the forced captcha.

Entries posted within the last 30 days can now receive up to the full 10k comments without a forced captcha. Entries older than 30 days keep the existing 5k threshold. All other captcha triggers (rate limiting, journal settings, spam heuristics) are unchanged.

Also refactors require_captcha_test to accept an LJ::Entry object instead of a bare ditemid, since the entry is already available at all call sites.

Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com

Commit: e78529d60c5e457534481f350c381821dc5810fd https://github.com/dreamwidth/dreamwidth/commit/e78529d60c5e457534481f350c381821dc5810fd Author: Mark Smith mark@dreamwidth.org Date: 2026-02-20 (Fri, 20 Feb 2026)

Changed paths: M .gitignore

Log Message:

Add .worktrees/ to .gitignore

Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com

Compare: https://github.com/dreamwidth/dreamwidth/compare/86a23512cf4a...e78529d60c5e

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: 86a23512cf4aca54a77dd38f26348c19a338d83d https://github.com/dreamwidth/dreamwidth/commit/86a23512cf4aca54a77dd38f26348c19a338d83d Author: Nova sda1@umbc.edu Date: 2026-02-20 (Fri, 20 Feb 2026)

Changed paths: M cgi-bin/LJ/S2/EntryPage.pm

Log Message:

Add per-entry OpenGraph meta tags for public entries (#3522)

Emit og:title, og:type, og:url, og:site_name, og:description, og:image, article:published_time, article:author, and article:tag for public journal entries so that Discord, Slack, Facebook, and other services can generate meaningful link previews.

Only public entries get these tags — non-public entries keep the existing site-wide defaults (DW logo) to avoid content leakage. All attribute values are escaped with LJ::ehtml(). The entry OG block is prepended to head_content so the entry-specific og:image takes precedence over the site-wide fallback from Page().

Fixes #2206

Co-authored-by: Novalinium nova@noblejury.com Co-authored-by: Claude Opus 4.6 noreply@anthropic.com

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: 722aca660143d17125cf5f95365e3d8272f5773c https://github.com/dreamwidth/dreamwidth/commit/722aca660143d17125cf5f95365e3d8272f5773c Author: Mark Smith mark@dreamwidth.org Date: 2026-02-20 (Fri, 20 Feb 2026)

Changed paths: M htdocs/js/jquery.shortcuts.nextentry.js

Log Message:

Fix k shortcut sometimes not scrolling to previous entry

prevPageEntry() used a strict < comparison with no tolerance, so if the scroll position landed even 1px past an entry's top (due to sub-pixel rounding or animation overshoot), pressing k would "scroll" to the entry already on screen, appearing to do nothing. Add the same 50px tolerance that nextPageEntry() already uses.

Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications

kaberett

Breakfast dal. This experiment continues to work extremely well.
I have definitely reached the point with the Incomplete White Puzzle where it's speeding up significantly on account of enough pieces are in place to significantly reduce the number of possible combinations that need checking. Today's decision was to start filling in from the bottom edge, where I still had a chunk that was just edge and no middles, because I think that up in the top left (interior) corner I've identified The Missing Piece, and will get annoyed if I wind up with non-contiguous gaps...
Today alternating Locate One Puzzle Piece with Do One Useful Job has been nice and smooth and easy. I have got Several things done. Is pleased.
Really really enjoying my ridiculous washi tape collection. Today I self-indulgently Added More Week Dividers, including replacing some pre-existing ones that I was Not Enjoying, Actually.
( Exercise & embodiment. )

kaberett

I have introduced my mother to this, I have introduced the Child's household to this, I am writing it down because clearly It Is Time for me to do so.

( Read more... )

wildeabandon

So the Church of England has drawn the "Living in Love and Faith" process to a close, in a way that puts any pursuit of my priestly vocation out of reach for the foreseeable future. A new working group is being set up to continue looking at the question of priests in same-sex marriages, which is supposed to report back to Synod in 2028. Based on past experience, that probably means 2029 or 2030, at which point there will no doubt be a new round of painful arguments, and then I guess we'll see. But for now, that door is closed.

I think I am currently feeling less upset about this than I thought I'd be, although it might just be alexithymia fogging things up. It didn't really come as a surprise, so to some extent letting go of the uncertainty is something of a relief.

It also removes the potential complication that comes with having reinvigorated my academic vocation, coming back to the field with my mental health intact, my ADHD treated, and the general increased wisdom that comes with age. Of course academia and the priesthood is hardly a combination that hasn't been tried before, but I had been worrying slightly about what happens if I have to make a choice about which to pursue first, and now that that choice has been taken off the table I can just concentrate on my studies, and should at least be well into a PhD before the question of formal priestly discernment becomes pertinent again.

kaberett

Specifically: I find myself in possession of both a superking duvet cover and a deep fitted double sheet that are mostly Genuinely Nice Cotton... and have both got holes worn through them in one specific place.

I have accepted about myself that I am not a person who will tolerate sleeping on patched bedsheets (because Textures). I am loathe to just hand them over to rag recycling. I am scared of trying to sew anything out of them, but might manage it with some encouragement.

I would greatly appreciate people Being Opinionated on this topic.

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: 472e7ca0671a6f37896549ef3e935b8cfdd52f03 https://github.com/dreamwidth/dreamwidth/commit/472e7ca0671a6f37896549ef3e935b8cfdd52f03 Author: Mark Smith mark@dreamwidth.org Date: 2026-02-16 (Mon, 16 Feb 2026)

Changed paths: M bin/worker/ses-incoming-email

Log Message:

Fix SES worker to expect S3 action notifications

The SNS notification now comes from the S3 action's topic_arn (terraform change), so receipt.action will always be type S3. Revert the non-S3 skip to a proper error since it would indicate a misconfiguration.

Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com

Commit: 4895912982e33afbf41b439bc7d38d1a625b25b7 https://github.com/dreamwidth/dreamwidth/commit/4895912982e33afbf41b439bc7d38d1a625b25b7 Author: Mark Smith mark@dreamwidth.org Date: 2026-02-17 (Tue, 17 Feb 2026)

Changed paths: M bin/worker/dw-embeds M cgi-bin/DW/Task.pm M cgi-bin/DW/TaskQueue.pm M cgi-bin/DW/TaskQueue/SQS.pm M t/taskqueue-dedup.t

Log Message:

Add max_retries support to TaskQueue to avoid DLQ pileup

Failed SQS tasks that exceed the receive count threshold are now marked complete instead of being left for the dead letter queue. The SQS layer passes ApproximateReceiveCount through to tasks, and start_work checks it against a configurable max_retries option. Applied to dw-embeds where unreachable remote hosts cause repeated failures that fill the DLQ.

Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com

Compare: https://github.com/dreamwidth/dreamwidth/compare/6792621c5547...4895912982e3

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications

kaberett

Yesterday afternoon I'd been discussing auditor traps. Yesterday evening we walked out of the supermarket and were confronted by

three signs across a blocked-off road: DIVERSION pointing right, ROAD CLOSED, DIVERSION pointing left

[description in alt text, better to follow]

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: 6792621c554719720fc25eba12b4ea612264cb73 https://github.com/dreamwidth/dreamwidth/commit/6792621c554719720fc25eba12b4ea612264cb73 Author: Mark Smith mark@dreamwidth.org Date: 2026-02-16 (Mon, 16 Feb 2026)

Changed paths: M bin/worker/ses-incoming-email M cgi-bin/DW/IncomingEmail.pm

Log Message:

Fix log levels in SES incoming email pipeline

Skip non-S3 action notifications at info level (expected duplicate from SNS action, not an error)
Empty email data is warn, not error
No support category match is info, not error (normal for non-support mail)

Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: def63d7f03f190bc1a99d4fd648fa0f1424b6960 https://github.com/dreamwidth/dreamwidth/commit/def63d7f03f190bc1a99d4fd648fa0f1424b6960 Author: Mark Smith mark@dreamwidth.org Date: 2026-02-16 (Mon, 16 Feb 2026)

Changed paths: A .github/workflows/tasks/worker-ses-incoming-email-service.json R .github/workflows/worker-build.yml R .github/workflows/worker-deploy.yml M .github/workflows/worker22-deploy.yml M config/update-workflows.py

Log Message:

Deprecate 18.04 workers

Also add deploy for new ses-incoming-email worker.

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: bba49ec2ee390e6882eda0b8004c6c57beacad9e https://github.com/dreamwidth/dreamwidth/commit/bba49ec2ee390e6882eda0b8004c6c57beacad9e Author: Mark Smith mark@dreamwidth.org Date: 2026-02-16 (Mon, 16 Feb 2026)

Changed paths: A bin/worker/ses-incoming-email A cgi-bin/DW/IncomingEmail.pm M cgi-bin/DW/Task/IncomingEmail.pm M config/workers.json A t/incoming-email.t

Log Message:

Add SES incoming email worker and shared processing pipeline

Extract email processing logic from DW::Task::IncomingEmail into DW::IncomingEmail so both the legacy TaskQueue worker and the new SES-based worker share the same pipeline. Remove obsolete spam filtering (subject heuristics, body checks, virus signatures, bounce detection) — SES handles this upstream now.

New components: - DW::IncomingEmail: shared processing (MIME parsing, hooks, post-by-email, alias forwarding with From-rewriting, support routing, multi-domain normalization via lj::INCOMING_EMAIL_DOMAINS) - bin/worker/ses-incoming-email: SQS-polling worker that receives SNS notifications from SES, fetches email from S3, checks SES spam/virus verdicts, and delegates to DW::IncomingEmail - t/incoming-email.t: tests for process flow, From-rewriting with per-sender SHA256 hash, and domain normalization

Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications

diffrentcolours

I've just applied to Companies House to have my Limited Company, of which I am sole director, struck off and dissolved. I founded it years ago as a potential holding organisation for providing IT services to political organisations. But it never really worked out like that - I spent many years providing those services such as mailing lists, chatrooms, wikis, CRM systems, but never got on the "approved suppliers" list. Then 2019 happened and... yeah, I stopped doing that. Part of people fucking me over in politics was making false allegations about me violating data protection laws, and that shit could have damaged my real job, so it was another reason to back away.

I'm feeling bittersweet about it all, but I still have my head held high. I did a better job than many of the people who actually got paid for doing this sort of thing. In retrospect I'm glad I never did manage to get anywhere with it; losing my social life and activism was bad enough, but at least I didn't lose my livelihood at the same time. Still, it's an interesting trouser-leg of time to contemplate.

I guess this is why I do things like read my employer's social impact reports. It's nice to know that the work I'm doing is helping other people make a difference in the world, even if it doesn't feel like I'm doing it myself.

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: 09b2fa09012b1edc5f23883029cf12a5a045bcc4 https://github.com/dreamwidth/dreamwidth/commit/09b2fa09012b1edc5f23883029cf12a5a045bcc4 Author: Mark Smith mark@dreamwidth.org Date: 2026-02-16 (Mon, 16 Feb 2026)

Changed paths: M config/workers.json

Log Message:

Add SQS queue names, so we can target autoscaling to queue depth

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: e18e34f910eb489a5d0daa40f93e2024cff9bbc2 https://github.com/dreamwidth/dreamwidth/commit/e18e34f910eb489a5d0daa40f93e2024cff9bbc2 Author: Mark Smith mark@dreamwidth.org Date: 2026-02-16 (Mon, 16 Feb 2026)

Changed paths: M .github/workflows/ci.yml M t/plack-auth.t M t/plack-integration.t M t/plack-media.t M t/plack-subdomain.t M t/plack-sysban.t

Log Message:

Expand CI to run plack, cleaner, and routing tests

Rename the workflow to "CI (fast)" and add plack integration, text cleaner, and routing test steps. Fix pre-existing test failures caused by the rate-limiting middleware interfering with unrelated tests: mock DW::RateLimit, session resolution, sysban, and uniq cookie in tests that don't exercise those concerns. Fix routing mock in sysban and subdomain tests to return 0 (handled) instead of undef (which fell through to BML). Suppress 'used only once' warnings across plack test files.

Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications

github posting in

changelog

Branch: refs/heads/main Home: https://github.com/dreamwidth/dreamwidth Commit: 7f5a36ba49391219cb773eeb2ee5248be3b9ac09 https://github.com/dreamwidth/dreamwidth/commit/7f5a36ba49391219cb773eeb2ee5248be3b9ac09 Author: Mark Smith mark@dreamwidth.org Date: 2026-02-16 (Mon, 16 Feb 2026)

Changed paths: M cgi-bin/Plack/Middleware/DW/Auth.pm M t/plack-auth.t

Log Message:

Fix domain session cookie bounce not firing under Plack

When a user's domain session cookie is stale (e.g. after re-login gives them a new session ID), the Plack Auth middleware detected the mismatch and set the bounce URL but never actually redirected. The page would render as logged-out instead of bouncing through /misc/get_domain_session to refresh the cookie.

Add the redirect check to the Auth middleware's else branch so all Plack-served pages (journal views, controllers, etc.) get the bounce. Also fix pre-existing test failures in plack-auth.t where MockUser was missing the userid method needed by the RateLimit middleware.

Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com

To unsubscribe from these emails, change your notification settings at https://github.com/dreamwidth/dreamwidth/settings/notifications