{"chain":[{"channel":"cities","content":" I appear to be heading south from Alberta.\r\n Great Falls, Montana, is a mid-sized city in north-central Montana along the Missouri River, known for its natural beauty and historical significance. The city is named after the series of waterfalls on the river, famously explored during the Lewis and Clark Expedition. Surrounded by mountain ranges like the Little Belt and Highwood Mountains, it offers abundant outdoor activities, including hiking, fishing, and camping. Local attractions include the Lewis and Clark Interpretive Center, Giant Springs State Park, and the Rivers Edge Trail. Great Falls also has a strong connection to Indigenous history and culture, a thriving agricultural economy, and a notable military presence through Malmstrom Air Force Base.\r\n\r\n----\r\n\r\nprojects for the week are just atacama. ( we cannot confirm-or-deny a *secret* project)\r\n\r\n----\r\n\r\natacama goals for the week:\r\n# clean up the \"tools for using claude\" code.\r\n# maybe finally add some LLM features\r\n# maybe try << Claude can't fix the parser >> again. ( there are various approaches here.)\r\n# maybe try << SMTP-in/IMAP-in/SMTP-out APIs >> again.\r\n# maybe continue to iterate on the \"channel security\" model\r\n# maybe ponder the many implications of a \"multi-user\" model.\r\n# maybe fix the \"quotes\" system\r\n# maybe do more \"put Claude through the paces by generating unit tests\"","created_at":"2025-01-20T18:37:06.215647","id":131,"is_target":false,"parent_id":null,"processed_content":"

\ud83d\udca1 I appear to be heading south from Alberta.\r

\n

\ud83e\udd16 Great Falls, Montana, is a mid-sized city in north-central Montana along the Missouri River, known for its natural beauty and historical significance. The city is named after the series of waterfalls on the river, famously explored during the Lewis and Clark Expedition. Surrounded by mountain ranges like the Little Belt and Highwood Mountains, it offers abundant outdoor activities, including hiking, fishing, and camping. Local attractions include the Lewis and Clark Interpretive Center, Giant Springs State Park, and the Rivers Edge Trail. Great Falls also has a strong connection to Indigenous history and culture, a thriving agricultural economy, and a notable military presence through Malmstrom Air Force Base.\r

projects for the week are just atacama. \ud83d\udd25( we cannot confirm-or-deny a secret project)\r

atacama goals for the week:\r

\n
    \n
  • clean up the \"tools for using claude\" code.\r
    • \n
  • maybe finally add some LLM features\r
    • \n
  • maybe try Claude can't fix the parser again. \ud83d\udca1( there are various approaches here.)\r
    • \n
  • maybe try SMTP-in/IMAP-in/SMTP-out APIs again.\r
    • \n
  • maybe continue to iterate on the \"channel security\" model\r
    • \n
  • maybe ponder the many implications of a \"multi-user\" model.\r
    • \n
  • maybe fix the \"quotes\" system\r
    • \n
  • maybe do more \"put Claude through the paces by generating unit tests\"
    • \n
","subject":"great falls (1/5)"},{"channel":"cities","content":"other than \"get Claude to write a regression test\", my attention was on projects other than Atacama yesterday.\r\n\r\n----\r\n\r\ntoday, considering 3 features:\r\n* multi-line quotes ( which would additionally weigh on the already-overloaded regex-based parser)\r\n* embedded images ( which raises the question: where are the images hosted?)\r\n* embedded YouTube videos ( probably as a \"television\" emoji that one clicks on to display the embedded viewer)\r\n\r\nthings like \"SMTP API\" simply aren't worth the *complexity*, much less my time to get the *machine* to implement it. ( the << compute cost >> of having the *machine* implement the SMTP API, presumably, is at most $1-2; it is not a relevant factor.) ( the price to me is already included in my monthly subscription. the \"cost\" is different than the \"price\")\r\n\r\nas far as the \"quotes\" system: it seems like something to leave alone. either let it rot, or wait for the improvements to become more evident.\r\n\r\nas far as the LLM features ... ( As one grows older, one sees the impossibility of imposing your will on the chaos with brute force. But, if you\u2019re patient, there may come the moment when, while you are eating an apple, the solution presents itself and says, \"Here I am.\" )","created_at":"2025-01-21T15:58:48.382850","id":139,"is_target":false,"parent_id":131,"processed_content":"

other than \"get Claude to write a regression test\", my attention was on projects other than Atacama yesterday.\r

today, considering 3 features:\r

\n
    \n
  • multi-line quotes \u2699\ufe0f( which would additionally weigh on the already-overloaded regex-based parser)\r
    • \n
  • embedded images \ud83d\udca1( which raises the question: where are the images hosted?)\r
    • \n
  • embedded YouTube videos \ud83d\udca1( probably as a \"television\" emoji that one clicks on to display the embedded viewer)\r
    • \n
\n

things like \"SMTP API\" simply aren't worth the complexity, much less my time to get the machine to implement it. \u2699\ufe0f( the compute cost of having the machine implement the SMTP API, presumably, is at most $1-2; it is not a relevant factor.) \ud83d\udca1( the price to me is already included in my monthly subscription. the \"cost\" is different than the \"price\")\r

\n

as far as the \"quotes\" system: it seems like something to leave alone. either let it rot, or wait for the improvements to become more evident.\r

\n

as far as the LLM features ... \ud83d\udcac( As one grows older, one sees the impossibility of imposing your will on the chaos with brute force. But, if you\u2019re patient, there may come the moment when, while you are eating an apple, the solution presents itself and says, \"Here I am.\" )

","subject":"great falls (2/5)"},{"channel":"cities","content":"things done on Atacama yesterday:\r\n# multi-line quotes\r\n# embedded YT player\r\n# make chess pieces easier to distinguish\r\n# clean up the \"regex\" parser\r\n# try again at the \"lexer\" parser ( if you chop a task up *enough*, maybe eventually Claude can handle it)\r\n\r\n----\r\n\r\nthings to-do today:\r\n# other projects in the morning\r\n# the \"lexer\" parser, again\r\n# probably more of \"other\" things after that","created_at":"2025-01-22T16:00:24.753079","id":140,"is_target":false,"parent_id":139,"processed_content":"

things done on Atacama yesterday:\r

\n
    \n
  • multi-line quotes\r
    • \n
  • embedded YT player\r
    • \n
  • make chess pieces easier to distinguish\r
    • \n
  • clean up the \"regex\" parser\r
    • \n
  • try again at the \"lexer\" parser \ud83d\udca1( if you chop a task up enough, maybe eventually Claude can handle it)\r
    • \n

things to-do today:\r

\n
    \n
  • other projects in the morning\r
    • \n
  • the \"lexer\" parser, again\r
    • \n
  • probably more of \"other\" things after that
    • \n
","subject":"great falls (3/5)"},{"channel":"cities","content":"the new \"lexer\"-based parser has crossed the hump of possibility.\r\n\r\nit took roughly 2 days of effort (plus 3 days of complaining, earlier). this is, arguably, slower than it would have been for me to write this *without* using the *machine*.\r\n\r\nthere are a few minor things to fix before it replaces the new system. those are in the next list.\r\n\r\n----\r\n\r\ntoday's things:\r\n# New lexer bugfixes - lists format incorrectly, PGN templates don't work, quotes aren't stored\r\n# Fix the test framework - occasionally it deletes the (local) prod database\r\n# Better export/import tools - should be able to load the prod database locally by transferring a JSON file\r\n# *Don't* try to add << i-Ching >> based features\r\n# Try to get the << LLM Magic Query >> working - based on a short prompt (\"Tell me about George Washington\", \"What does baking powder do\"), figure out the best extended prompt, and then get the response.","created_at":"2025-01-23T15:34:38.768053","id":145,"is_target":false,"parent_id":140,"processed_content":"

the new \"lexer\"-based parser has crossed the hump of possibility.\r

\n

it took roughly 2 days of effort (plus 3 days of complaining, earlier). this is, arguably, slower than it would have been for me to write this without using the machine.\r

\n

there are a few minor things to fix before it replaces the new system. those are in the next list.\r

today's things:\r

\n
    \n
  • New lexer bugfixes - lists format incorrectly, PGN templates don't work, quotes aren't stored\r
    • \n
  • Fix the test framework - occasionally it deletes the (local) prod database\r
    • \n
  • Better export/import tools - should be able to load the prod database locally by transferring a JSON file\r
    • \n
  • Don't try to add i-Ching based features\r
    • \n
  • Try to get the LLM Magic Query working - based on a short prompt (\"Tell me about George Washington\", \"What does baking powder do\"), figure out the best extended prompt, and then get the response.
    • \n
","subject":"great falls (4/5)"},{"channel":"cities","content":"yesterday:\r\n* fix a lot of bugs in the \"lexer\" parser (pinyin, quotes, parenthesis, Chinese, line-breaks)\r\n* new helper-scripts (pygrep, interactive shell default)\r\n* fix the \"tests access the real database\" bug\r\n\r\n----\r\n\r\ntoday:\r\n* maybe finally the LLM features?\r\n* but first, a \"taskqueue\" setup. I can't expect the *machine* to finish within 1 minute. ( or, perhaps this is the *nudge* to switch to a \"publishing\" model. where posts are written, and then, in a separate process, converted to web-visible after some improvements)\r\n\r\n----\r\n\r\nother news:\r\n* it was << Pretzel Day >> yesterday at the coworking space\r\n* my knees were sore this morning, in a different way. it felt like something between a twitch, neuropathy, pinched nerve, etc. it (mostly) went away after a half-hour. but I am still watching out for it.","created_at":"2025-01-24T16:25:21.953835","id":148,"is_target":true,"parent_id":145,"processed_content":"

yesterday:\r

\n
    \n
  • fix a lot of bugs in the \"lexer\" parser (pinyin, quotes, parenthesis, Chinese, line-breaks)\r
    • \n
  • new helper-scripts (pygrep, interactive shell default)\r
    • \n
  • fix the \"tests access the real database\" bug\r
    • \n

today:\r

\n
    \n
  • maybe finally the LLM features?\r
    • \n
  • but first, a \"taskqueue\" setup. I can't expect the machine to finish within 1 minute. \ud83d\udca1( or, perhaps this is the nudge to switch to a \"publishing\" model. where posts are written, and then, in a separate process, converted to web-visible after some improvements)\r
    • \n

other news:\r

\n
    \n
  • it was Pretzel Day yesterday at the coworking space\r
    • \n
  • my knees were sore this morning, in a different way. it felt like something between a twitch, neuropathy, pinched nerve, etc. it (mostly) went away after a half-hour. but I am still watching out for it.
    • \n
","subject":"great falls (5/5)"}]}