AUIS revival

Timeline
Login

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

50 most recent check-ins

2026-08-19
05:23
Add a png object Leaf check-in: 928a5ddc1c user: wdc tags: html
04:10
design doc: record remote image fetching as done, PNG decode as in-progress (concurrent instance) check-in: 8410b362b3 user: wdc tags: html
04:07
htmlatk.c: sniff real image format when declared Content-Type's decoder fails; suppress routine gif__Load stderr on the speculative first attempt; htmlatktest.c: fix binary-corrupting UTF-8 conversion in cid: test resolver check-in: 15c117b15d user: wdc tags: html
04:06
messages: fetch remote <img> content via curl (httpimg.c), gated by ams.loadremoteimages pref + per-message override, im.c SIGCHLD race fix, per-render-pass fetch cache check-in: 007ce9ec43 user: wdc tags: html
2026-08-18
21:46
design doc: record scrollbar endzone-blanking fix; note remaining paging-precision imprecision is separate, still open check-in: b3bf7ddd5a user: wdc tags: html
21:44
textv.c: fix BackSpace fast-path applied to a held scrollbar bottom-endzone click, blanking the screen check-in: 492e9b6e24 user: wdc tags: html
21:05
design doc: record MoveForward fix, note scrollbar endzone as next open bug check-in: 54ba011a57 user: wdc tags: html
21:05
textv: fix MoveForward blanking screen at end-of-document check-in: de73971e25 user: wdc tags: html
17:32
design doc: add Implementation section inventorying lset/lpair/textview fixes check-in: 31b432b585 user: wdc tags: html
17:13
textv.c: fix Visible() viewport check and BackSpace landing inside oversized lines check-in: 1971a51eb3 user: wdc tags: html
15:18
textv.c: fix BackSpace fast-path that no-op'd backward scroll through oversized lines check-in: 447f0cbaa7 user: wdc tags: html
04:57
htmlatk.c: BuildLsetCell splice-through only inlines true single-cell nested rows check-in: a9a82d66f1 user: wdc tags: html
04:38
porting-assessment.md: document the lset/lpair-as-HTML-table-layout-engine bug family (§24) -- width= allowlist gap, fixed-pixel splits, the 2048px and HORIZONTAL-branch DesiredSize bugs (all fixed), plus the still-open row/cell splice-through flattening found live-testing them check-in: db50e4f07a user: wdc tags: html
04:32
lpair.c: fix lpair__DesiredSize's HORIZONTAL (stacked) branch to sum children's real heights instead of echoing the offered height back -- caused a too-tall embedded HTML table to under-report its own height to textview's line layout, so <space> paging jumped straight past unread content instead of stopping partway through it (National Grid mail footer); found live-testing the 2048px-clamp fix from the prior commit against real mail check-in: b38e9fda90 user: wdc tags: html
03:40
htmlpart.c/lsetv/lpair/htmlatk.c: add width= to table/td/th attr_allowlist and give lset fixed-pixel splits (lsetview_MakeHorzFixed/VertFixed, lpair_TOPFIXED) for small decoration cells; fix lpair__DesiredSize's 2048px height clamp that squashed a whole row to a decoration's tiny height once real flowed content passed 2048px -- root cause of Book Rack still rendering blank after the width fix, live-verified fixed on Book Rack and National Grid check-in: c5f535ca9b user: wdc tags: html
2026-08-17
06:09
mimepart.c/text822.c: default multipart/alternative selection to text/html over text/plain (real newsletter mail is mostly this shape); add shared 'ams.preferplaintext' profile switch to restore the old order, classless (getprofileswitch, not environ_GetProfileSwitch) so mimepart.c stays linkable into cui, which has no Class system; options.c gets a new OT_PROFSWITCH toggle in messages' Set Options panel; messages.c adds a per-message 'Show as Plain Text' command (MODE822_HTMLPLAINTEXT, text822.ch) as an escape hatch for HTML mail whose deeply-nested table layout Stage 3 renders badly (Book Rack newsletter, 87 lset levels); mime-display-tests updated for the new default check-in: 492ac9cff2 user: wdc tags: html
04:43
mimepart.c/htmlpart.c: ASCII-fallback a small set of Unicode smart punctuation (curly quotes, en/em dash, ellipsis) instead of '?' -- both the raw-UTF-8 path (mimepart_Utf8ToLatin1) and the HTML-entity path (&rsquo;/&mdash;/etc, htmlpart.c); confirmed live, National Grid; design doc documents the fix plus a full script-coverage inventory (Cyrillic/Hebrew/Arabic/Bengali/CJK remain structurally unfixable under ATK's 8-bit Latin-1 model) check-in: 8a247e867b user: wdc tags: html
04:25
html-mail-rendering-design.md: document three gaps found live 2026-08-17 (smart-punctuation ?, unwired link clicks, unimplemented image resolution) with agreed sequencing; correct the stale Links/Remote-content sections to reflect what's actually wired vs. just planned check-in: 0dc9ec0a52 user: wdc tags: html
04:04
roadmap.md: bring HTML mail rendering project status up to date -- Stages 1-3 done and wired in, table/spread->lset/lpair pivot, endzone scrollbar bug logged, milestone entry added check-in: c0e80b78b7 user: wdc tags: html
04:00
revival/doc: document the isolated-blank-row fix + why it's still the right call, and log the endzone-click scrollbar bug (deliberately deferred, not HTML-specific) in revival.md's Open issues check-in: f66912e34a user: wdc tags: html
04:00
htmlatk.c/htmltext.c/htmlpart.c: drop every blank spacer <tr> (not just consecutive runs of 2+), and treat display:none/visibility:hidden as real hiding, not just cosmetic -- fixes 'blank line between every lset' and a hidden marketing preheader rendering as visible text, both found live against a real National Grid message (now saved at revival/tests/national-grid.html); htmlatktest.c gets DumpLsetTree + roundtrip verification support check-in: 1b3ea48cab user: wdc tags: html
03:54
lsetv.c/.ch: lsetview had no DesiredSize override, so a leaf silently ignored its real child content size (inherited from lpair's split-partitioning fallback); add DesiredSize forwarding + WantNewSize escalation to the real parent chain (modeled on celview's same fix), both needed for HTML-table-as-lset rows to auto-size correctly check-in: b2e1fed841 user: wdc tags: html
03:54
textv.c: BackSpace's EOF line-padding wrongly snapped a just-reached end-of-doc position back to the previous (tall) line, causing an infinite ^V scroll loop when the last line is a disproportionately tall embedded view -- fix scoped to the exact units==0 boundary case check-in: 0166d067de user: wdc tags: html
03:14
buffer.c+bufferlist.c: filter CRLF to LF on datastream open via crlffilter.h funopen() wrapper -- LF-only .ez parsers (lset.c etc.) desync on \r\n, warn once to stderr on normalize Leaf check-in: c3d1216d8c user: wdc tags: trunk
2026-08-16
05:40
revival/doc: table-cell/color-normalization design updates, writeds/roundtrip real-datastream verification technique documented in sonnet-playbook.md, lset-table-reflow follow-up prompt (table/spread is a fixed-pixel grid, doesn't reflow -- lset/lpair is percentage-based and does); render_test.ez is wdc's live verification artifact, regenerated across today's fixes check-in: 9d1f53bc7e user: wdc tags: html
05:39
htmlpart.c: drop zero-width Unicode entities (zwnj/zwj/lrm/rlm, named and numeric) silently instead of passing through literal text -- fixes 'hidden preheader' anti-clipping padding rendering as visible garbage in real mail check-in: d5361beffe user: wdc tags: html
05:39
text822.c+cui.c+Imakefile: wire Stage 3 renderer into messages/cui (Gate 5), retire mimepart_HtmlToText shim in both -- text822.c gets htmlatk_Render()+two-level fallback and a UTF-8-to-Latin1 conversion step it was missing (found live in real mail), cui.c gets the plain Stage 1/2 sequence only (no ATK dependency) check-in: 97c5f0ebc3 user: wdc tags: html
05:39
src/atkams/messages/lib/htmlatk.c+.h+htmlatktest.c: Stage 3 ATK-styled HTML renderer (Gates 1-4) plus same-day hardening -- recursive per-cell text/table embedding grounded in real Sherman.Alloc precedent (not the earlier text-vs-image dispatch mistake), CSS color normalization + a mark-count corruption fix it exposed, real column widths from HTML width= instead of ATK's 99px default, writeds/roundtrip test subcommands for real-datastream/ez verification check-in: 05de15e4aa user: wdc tags: html
00:38
revival/doc/html-atk-renderer-prompt.md: Stage 3 (ATK-styled HTML renderer) gated delegation prompt -- research gate for the unfamiliar ATK mechanics, then core styled text, table embedding, image embedding + clickable links, text822.c wiring check-in: 9fc6e1c33c user: wdc tags: html
2026-08-15
22:56
htmltext.c: Stage 2 of HTML mail rendering -- plain-text renderer walking htmlpart.c's tree (tables degrade to sequential text, links show text+url, images show alt-text placeholders); collapse Latin-1 NBSP as ordinary whitespace, not just ASCII space/tab/newline; revival/tests/html-totext-tests, 13/13 passing check-in: 56f4878403 user: wdc tags: html
22:56
give html/ its own build tree: repoint site.h's DEFAULT_ANDREWDIR_ENV, regenerate machdep Makefiles; document the bootstrap procedure + merge-back site.h exclusion in sonnet-playbook.md check-in: 443347e5a7 user: wdc tags: html
22:00
htmlpart.c: Stage 1 of HTML mail rendering -- ATK-independent tokenizer + sanitization-allowlist tree builder, modeled on mimepart.c; iterative (no C recursion) per the no-depth-cap design decision; revival/tests/html-parse-tests, 20/20 passing check-in: 756e137aa6 user: wdc tags: html
2026-08-14
07:30
revival/tests/html-fixtures: real-world HTML mail corpus (16 fixtures from the mirrored INBOX) for the html-mail-rendering project, with README/findings check-in: a6d9d49492 user: wdc tags: html
02:04
html-mail-rendering-design.md: drop fixed size/depth caps (real mail routinely exceeds any guessed threshold; dataobj.c's own recursive begindata reader has never needed one), key the renderer fallback on parse time instead; decide PNG support given jpeg/tif/raster precedent check-in: 44a2e8c098 user: wdc tags: html
2026-08-13
21:54
html branch: add html-mail-rendering-design.md (sanitization allowlist, table-object strategy, image capability-check strategy, explicit fallback contract); link from roadmap.md check-in: d3a7096472 user: wdc tags: html
21:45
Create new branch named "html" check-in: 1689674256 user: wdc tags: html
04:38
porting-changelog.md: log the andrew-6.4-into-trunk merge as the milestone it is check-in: 520e3672e6 user: wdc tags: trunk
04:29
sonnet-playbook.md/rollout-procedure.md: trunk is canon for debug sessions; add project-branch check-in flow for larger initiatives check-in: 1bcf127bd3 user: wdc tags: trunk
2026-08-12
21:30
quickstart.md: document config/Makefile's BASEDIR as a second hardcoded path alongside site.h (reported by an outside builder) check-in: 4379605690 user: wdc tags: trunk
21:25
quickstart.md/mail-quickstart.md: drop andrew-6.4-specific paths now that trunk is primary; add write-andrewsetup step for messages/cui check-in: 5d87cc06a7 user: wdc tags: trunk
21:24
Merge andrew-6.4 into trunk: revival docs, ANSI/LP64 fixes, restored adew/Arb, README for GitHub mirror check-in: 9e6cd5ec27 user: wdc tags: trunk
20:29
adew/Imakefile: build generated arb launcher under a collision-proof name; doc the Arb-clobber bug in revival.md/roadmap.md/porting-changelog.md Leaf check-in: 4c077a7c36 user: wdc tags: andrew-6.4
20:21
Restore src/atk/adew/Arb (was clobbered by generated arb launcher, dc4206084d) check-in: 2d016415ac user: wdc tags: andrew-6.4
04:54
claude-history: retire strict-prototypes-census.md, repoint references check-in: 19e71bac1f user: wdc tags: andrew-6.4
04:49
claude-history: retire roadmap-old.md/m4-rollout-runbook.md/m4-batches.md (repoint references tree-wide, add M4 README section); sonnet-playbook.md: document static-vs-dynamic runapp relink recipe check-in: 12119e063d user: wdc tags: andrew-6.4
04:49
ams-IMAP-project.md: repoint roadmap-old.md citation to claude-history/roadmap-old.md check-in: c2ddf652d6 user: wdc tags: andrew-6.4
04:45
Add an integral sign to an eq. check-in: 7138e28f66 user: wdc tags: andrew-6.4
03:16
porting-assessment.md/roadmap.md: correct eq integral-symbol misunderstanding, log real eq multi-line cleanup bug, mark §2/3/4/6/10/11/12/14/17/18/21 resolved, relocate build-up-strategy section, move status labels out of headings into the body check-in: b055ed0073 user: wdc tags: andrew-6.4
2026-08-11
21:23
Close out menu-delay investigation: confirm SkyLight/CoreAnimation wait via timestamp correlation, rule out frontmost-app theory, drop stale debug dump check-in: a424ea7c27 user: wdc tags: andrew-6.4
20:51
Add wall-clock timestamps to mdbg() menu-delay tracing for correlation with process sampling check-in: e1edebc412 user: wdc tags: andrew-6.4