jjt
@jjt.io
dad of two, software engineering leader of many
he/him, 🇨🇦
he/him, 🇨🇦
i think a very small, dumb part of my brain thinks switching to fish requires me to delete every posix-compliant shell from my machine, and thus i haven't tried yet
mise rules tho, replaced many version managers and direnv (so far)
mise rules tho, replaced many version managers and direnv (so far)
December 10, 2025 at 4:50 PM
i think a very small, dumb part of my brain thinks switching to fish requires me to delete every posix-compliant shell from my machine, and thus i haven't tried yet
mise rules tho, replaced many version managers and direnv (so far)
mise rules tho, replaced many version managers and direnv (so far)
yup!
1. listing bookmarks locally (cleaning old ones out, etc)
2. viewing them on github (github.com/<ORG>/<REPO>/branches)
3. telling someone a branch name verbally - they're more likely to remember "refactor-foo" rather than "jjt/push-xuniuir" :)
1. listing bookmarks locally (cleaning old ones out, etc)
2. viewing them on github (github.com/<ORG>/<REPO>/branches)
3. telling someone a branch name verbally - they're more likely to remember "refactor-foo" rather than "jjt/push-xuniuir" :)
github.com
October 26, 2025 at 5:10 PM
yup!
1. listing bookmarks locally (cleaning old ones out, etc)
2. viewing them on github (github.com/<ORG>/<REPO>/branches)
3. telling someone a branch name verbally - they're more likely to remember "refactor-foo" rather than "jjt/push-xuniuir" :)
1. listing bookmarks locally (cleaning old ones out, etc)
2. viewing them on github (github.com/<ORG>/<REPO>/branches)
3. telling someone a branch name verbally - they're more likely to remember "refactor-foo" rather than "jjt/push-xuniuir" :)
October 26, 2025 at 3:20 PM
in isolation i expect a good GUI will be superior than a TUI for 99.9% of use cases
jj-fzf isn't a type of tool i use, but ppl seem to like them (as they do lazygit) - maybe they value staying in the terminal more than a superior experience of a GUI?
idk, could just be novelty?
jj-fzf isn't a type of tool i use, but ppl seem to like them (as they do lazygit) - maybe they value staying in the terminal more than a superior experience of a GUI?
idk, could just be novelty?
September 21, 2025 at 8:48 PM
in isolation i expect a good GUI will be superior than a TUI for 99.9% of use cases
jj-fzf isn't a type of tool i use, but ppl seem to like them (as they do lazygit) - maybe they value staying in the terminal more than a superior experience of a GUI?
idk, could just be novelty?
jj-fzf isn't a type of tool i use, but ppl seem to like them (as they do lazygit) - maybe they value staying in the terminal more than a superior experience of a GUI?
idk, could just be novelty?
agreed, half-assed GUIs abound
even optimal GUIs have disadvantages over TUIs when context is considered: staying in terminal, getting stdout/err, look/feel of terminal command
ex. of latter is github.com/tim-janik/jj... which prints mutative jj commands to "help users in learning the jj CLI"
even optimal GUIs have disadvantages over TUIs when context is considered: staying in terminal, getting stdout/err, look/feel of terminal command
ex. of latter is github.com/tim-janik/jj... which prints mutative jj commands to "help users in learning the jj CLI"
September 21, 2025 at 8:46 PM
agreed, half-assed GUIs abound
even optimal GUIs have disadvantages over TUIs when context is considered: staying in terminal, getting stdout/err, look/feel of terminal command
ex. of latter is github.com/tim-janik/jj... which prints mutative jj commands to "help users in learning the jj CLI"
even optimal GUIs have disadvantages over TUIs when context is considered: staying in terminal, getting stdout/err, look/feel of terminal command
ex. of latter is github.com/tim-janik/jj... which prints mutative jj commands to "help users in learning the jj CLI"
i don’t use lazygit, but something beyond bare commands & still in terminal can be nice, and TUI is a forcing function for good keyboard controls compared to GUIs
September 21, 2025 at 3:45 AM
i don’t use lazygit, but something beyond bare commands & still in terminal can be nice, and TUI is a forcing function for good keyboard controls compared to GUIs
100% and i’d add that you never listen to music on speakers, you listen to it on the acoustic system of the speakers + the room together
room treatment if feasible is worth it, at least first reflection points
room treatment if feasible is worth it, at least first reflection points
September 18, 2025 at 3:09 PM
100% and i’d add that you never listen to music on speakers, you listen to it on the acoustic system of the speakers + the room together
room treatment if feasible is worth it, at least first reflection points
room treatment if feasible is worth it, at least first reflection points
and top-level change (id: wozmn) just points to tip of the evolog stack (commit id: 79500)
sidenote: they use the end of base36 for change ids to disambiguate the two, so commit digits are [0-f] (like git) but change ids are [z-k]
sidenote: they use the end of base36 for change ids to disambiguate the two, so commit digits are [0-f] (like git) but change ids are [z-k]
September 10, 2025 at 6:01 PM
and top-level change (id: wozmn) just points to tip of the evolog stack (commit id: 79500)
sidenote: they use the end of base36 for change ids to disambiguate the two, so commit digits are [0-f] (like git) but change ids are [z-k]
sidenote: they use the end of base36 for change ids to disambiguate the two, so commit digits are [0-f] (like git) but change ids are [z-k]
i've only had to go to the evolog a few times, but valuable to have that history of how a commit evolved over time
like in git if you wanted to see a change between last git commit and current state in working tree/index
like in git if you wanted to see a change between last git commit and current state in working tree/index
September 10, 2025 at 5:55 PM
i've only had to go to the evolog a few times, but valuable to have that history of how a commit evolved over time
like in git if you wanted to see a change between last git commit and current state in working tree/index
like in git if you wanted to see a change between last git commit and current state in working tree/index
worth it, imo
evolog is neat, like a mini branch for a given commit/change
i changed a file 3 times, evolog shows commits from jj snapshotting from that (usually happens w/ jj cmd invocation, i have watchman integration which does it every file change)
rich history without cruft in top-level
evolog is neat, like a mini branch for a given commit/change
i changed a file 3 times, evolog shows commits from jj snapshotting from that (usually happens w/ jj cmd invocation, i have watchman integration which does it every file change)
rich history without cruft in top-level
September 10, 2025 at 5:51 PM
worth it, imo
evolog is neat, like a mini branch for a given commit/change
i changed a file 3 times, evolog shows commits from jj snapshotting from that (usually happens w/ jj cmd invocation, i have watchman integration which does it every file change)
rich history without cruft in top-level
evolog is neat, like a mini branch for a given commit/change
i changed a file 3 times, evolog shows commits from jj snapshotting from that (usually happens w/ jj cmd invocation, i have watchman integration which does it every file change)
rich history without cruft in top-level
yup, daily driver for 1yr now, just a better tool all around
op log for repo history (jj undo/redo), working copy is a commit instead of index (no stash!), rebases always succeed, great revset language, lots to love
some rough edges here and there but colocated jj repos can use git interchangeably
op log for repo history (jj undo/redo), working copy is a commit instead of index (no stash!), rebases always succeed, great revset language, lots to love
some rough edges here and there but colocated jj repos can use git interchangeably
September 10, 2025 at 2:23 AM
yup, daily driver for 1yr now, just a better tool all around
op log for repo history (jj undo/redo), working copy is a commit instead of index (no stash!), rebases always succeed, great revset language, lots to love
some rough edges here and there but colocated jj repos can use git interchangeably
op log for repo history (jj undo/redo), working copy is a commit instead of index (no stash!), rebases always succeed, great revset language, lots to love
some rough edges here and there but colocated jj repos can use git interchangeably
np! i've been a very happy jj user for a while now, due to its refined and simpler mental models as you put it - worth the rough edges for me
the discord is also full of helpful and welcoming people discord.gg/PUDv9rtT
the discord is also full of helpful and welcoming people discord.gg/PUDv9rtT
July 24, 2025 at 5:21 PM
np! i've been a very happy jj user for a while now, due to its refined and simpler mental models as you put it - worth the rough edges for me
the discord is also full of helpful and welcoming people discord.gg/PUDv9rtT
the discord is also full of helpful and welcoming people discord.gg/PUDv9rtT
there's the experimental `advance-branches` feature github.com/jj-vcs/jj/di..., but i and many ppl define a `tug` alias github.com/jj-vcs/jj/di...
auto-local-bookmark config might replicate `git pull`? jj-vcs.github.io/jj/latest/co...
auto-local-bookmark config might replicate `git pull`? jj-vcs.github.io/jj/latest/co...
July 24, 2025 at 5:16 PM
there's the experimental `advance-branches` feature github.com/jj-vcs/jj/di..., but i and many ppl define a `tug` alias github.com/jj-vcs/jj/di...
auto-local-bookmark config might replicate `git pull`? jj-vcs.github.io/jj/latest/co...
auto-local-bookmark config might replicate `git pull`? jj-vcs.github.io/jj/latest/co...
vim guy in a group chat switched full time to hx a few mos ago, he loves it, speaks highly of its minor modes and how put together everything is
Keymap
docs.helix-editor.com
July 3, 2025 at 5:22 PM
vim guy in a group chat switched full time to hx a few mos ago, he loves it, speaks highly of its minor modes and how put together everything is
been on (n+)vim for almost 2 decades now, and selection-first is likely better UX than action-first
i love flash.nvim in treesitter mode to do visual selection then act
i love flash.nvim in treesitter mode to do visual selection then act
July 3, 2025 at 5:17 PM
been on (n+)vim for almost 2 decades now, and selection-first is likely better UX than action-first
i love flash.nvim in treesitter mode to do visual selection then act
i love flash.nvim in treesitter mode to do visual selection then act
i think about the hex grid article often, esp the orientation of the hexes (pointy up vs flat up) and how it percolates through every example and formula
June 8, 2025 at 4:50 PM
i think about the hex grid article often, esp the orientation of the hexes (pointy up vs flat up) and how it percolates through every example and formula
risks further entrenching incumbent PLs, frameworks, tools, libraries, etc
effect was there before (pick popular, stable, widespread tech) just accelerated now
effect was there before (pick popular, stable, widespread tech) just accelerated now
May 6, 2025 at 4:20 PM
risks further entrenching incumbent PLs, frameworks, tools, libraries, etc
effect was there before (pick popular, stable, widespread tech) just accelerated now
effect was there before (pick popular, stable, widespread tech) just accelerated now
but likely need a fair and unbiased media to do so - at least polls are somewhat more objective than editorials and slants in news stories
April 26, 2025 at 7:25 PM
but likely need a fair and unbiased media to do so - at least polls are somewhat more objective than editorials and slants in news stories
yup, no way to measure and publicly report on the system without affecting it, defeating the purpose of that measurement
i often wonder if democracies would be better off disallowing polls
i often wonder if democracies would be better off disallowing polls
April 26, 2025 at 7:23 PM
yup, no way to measure and publicly report on the system without affecting it, defeating the purpose of that measurement
i often wonder if democracies would be better off disallowing polls
i often wonder if democracies would be better off disallowing polls