| Age | Commit message (Collapse) | Author |
|---|
|
When building per-module derivations, copying dependency .hi files
was failing silently because: 1. First dep copies read-only Alpha.hi
from nix store 2. Second dep also has Alpha.hi (transitive dep) and
tries to overwrite 3. cp fails with permission denied but error is
hidden by '|| true'
Fixed by: - Add -f flag to cp to force overwrite - Run chmod -R +w
after each dependency copy - This allows later deps to overwrite
shared files like Alpha.hi
Tested: Biz/Que/Host.hs now builds successfully with per-module
derivations
|
|
Set dontPatchShebangs=true for per-module link derivation to avoid
unbound variable error in Nix's patch-shebangs.sh script.
The script declares 'local update' but never initializes it, causing
bash set -u to fail. Our ELF binaries don't need shebangs patched
anyway.
Tested: All Haskell targets now build successfully including
Omni/Lint.hs which uses makeWrapper with rundeps.
|
|
- Fix nixBuild to propagate exit code from realise step
- Previously used >> which discarded exit code - Now checks realise
result before running symlink - Fixes false success checkmarks on
build failures
- Fix per-module Nix derivations
- Use cp -rL instead of tar (src is directory not tarball) - Add
coreutils and findutils to pkgs - Copy deps to . and use -i. for
GHC - Use find with --parents to preserve module hierarchy - Set
dontStrip=true to avoid fixup script errors
- Tested: Example.hs, Task.hs, Dragons.hs, Bild.hs all build - Known
issue: makeWrapper fixup scripts have unbound vars (Nix bug)
|
|
When using custom builder, unpackPhase function is not available.
Use tar xzf to manually extract the source archive.
|
|
When using custom builder, standard phases don't run automatically.
Call unpackPhase explicitly and cd to 'source' directory where files
are unpacked.
Fixes 'sourceRoot: unbound variable' error in module compilation.
|
|
The haskell.ghcWith package has setup hooks that override buildPhase
even when explicitly set. Solution: use custom builder = stdenv.shell
with args instead of relying on mkDerivation phases.
Changes: - Module derivations: Use custom builder with single -c script
- Combines unpack, build, and install into one script - Explicitly
call ghc with full path to avoid hook interference - Remove unused
objectPaths computation (now using ghc --make with source)
This fixes builds for Omni/Lint.hs, Omni/Task.hs, and all other
Haskell targets with complex dependency graphs.
|
|
GHC --make was treating -i /nix/store/.../hidir paths as compilation targets.
Solution: Copy all .hi files to local directory and use -i. only.
Also make hsGraph optional with 'or null' for backward compatibility with
old bild binaries during bootstrap.
Amp-Thread-ID: https://ampcode.com/threads/T-fe68faaf-1c1d-4c43-a377-1cf5e6cffb3a
Co-authored-by: Amp <amp@ampcode.com>
|
|
GHC --make interprets arguments before flags as targets. Moving the
entry point source file to the beginning prevents -i paths from being
treated as compilation targets.
This fixes the 'is not a module name or a source file' error.
|
|
This is the core architecture transformation from Phase 3 of the
performance plan. Each Haskell module is now built as a separate
Nix derivation, enabling true incremental builds where only changed
modules and their dependents are rebuilt.
Implementation: - buildHsModuleGraph: Analyzes transitive module
dependencies and builds DAG - TH detection: Falls back to monolithic
build if Template Haskell detected - SCC cycle detection: Falls
back if import cycles found - Per-module Nix builder: Each module ->
separate derivation with .hi and .o - Module dependencies: Copy .hi
files to build dir, use -i flags for imports - Final link: Use ghc
--make with entry point source + -i paths to .hi files - Entry point
fix: Explicitly analyze entry point module separately from deps
Architecture: - Module compilation: ghc -c with -i paths to dependency
.hi files - Source filtering: Each module derivation includes only
its source file - Dependency DAG: Expressed as recursive Nix attrset
with lib.fix - Link phase: ghc --make with entry source file + all .hi
search paths - Fallback: Monolithic ghc --make when hsGraph is null
(TH/cycles)
Performance characteristics: - Change one module -> rebuild only
that + dependents + relink - Nix handles DAG scheduling and caching
automatically - Parallel module compilation (Nix orchestrates) -
Content-addressed caching across machines
Testing: - Added test_buildHsModuleGraph unit test - Verified with
Omni/Bild/Example.hs (4 modules) - Tested incremental rebuild triggers
correct subset
This completes Phase 2 and Phase 3 core optimizations from the plan.
|
|
Only include directories that are ancestors of source files in
allSources. Previously accepted all directories, causing rebuilds
when any new directory was added to the repo.
Implementation: - Precompute normalized source paths and their ancestor
directories - Filter directories against allowedDirs whitelist -
Normalize paths in file filter for consistency - Keep existing skip
list behavior for _ and .direnv
This is optimization #2 from Phase 2 of the performance plan.
|
|
I think the calls to Log.setup() were accidentally creating multiple
loggers, hopefully this fixes the problem.
|
|
- Add stripe to Omni/Bild/Deps/Python.nix (alphabetically sorted) -
Fix all type annotations in Billing.py for mypy - Document how to add
Python packages in AGENTS.md - Add billing routes to Web.py (checkout,
portal, webhook)
This enables Stripe integration in PodcastItLater.
Related to task t-144e7lF
|
|
Check to prevent processing of large articles, truncate oversized
content, defer jobs during high memory usage, use streaming TTS
generation and upload to minimize memory consumption.
|
|
According to Claude, the nixos-unstable branch has more stuff in it's binary
cache than nixos-unstable-small. Idk, let's try it and find out.
|
|
This implements a working prototype of PodcastItLater. It basically just works
for a single user currently, but the articles are nice to listen to and this is
something that we can start to build with.
|
|
This is actually useful because nix only shows the last 25 log lines, so I'd
like to know how deep the problems go with a summary.
|
|
The colored output is not decoded by the nix log accumulator, so you just get
these control characters. Get rid of them for more readable output.
|
|
This took quite a bit of work to package, and it's not a perfect package, but I
think it will do to get Biz/Astrologer.py working.
|
|
I had to update nixos-unstable-small and nixos-24_11 and disable a bunch of
tests, and then re-compile like the whole world, but aider is actually working,
finally.
|
|
Also improves the Example.py a bit with a real test case and example of how to
use Omni.Test.
|
|
Pydantic-ai is an agent framework that seems simple and good: well-typed with
pydantic, tool usage is just an `@tool` decorator on a function, and so
on. While building these I realized there were some deps they needed that were
already in nixpkgs unstable, so I just switched to that instead of trying to
backport all the versions and stuff.
|
|
This does the full transition: web server, mail server, xmpp. I expect some
disruption, but hopefully this is everything and it just switches over without
any problem.
|
|
These are broken in upstream nixpkgs. Updating didn't fix them, so I'm just
gonna disable them for now. I'm using pipx to install aider anyway.
Also: Remove _ literal from GLOBIGNORE. I think it was causing the **/* glob to
include stuff in the cabdir, idk why.
|
|
This is just so I can run gemma3. Ollama has more than just a src hash, it needs
the godeps hash as well, so I can't simply track it with `deps` unfortunately.
|
|
Some things got through the CI system, probably during updates. So I had more
stupid little fixes to do. I should really improve bild to the point that these
won't happen anymore and builds don't take so long.
|
|
I got this from the upstream branch jd/zx2c4-deployment, which does have the
below commit implementing dark mode. I first cherry-picked the commit to master
but it wouldn't build. Then I realized that the last cut release was v1.2.3 from
5 years ago, and that's what nixpkgs has pinned, so I cherry-picked it onto
v1.2.3 and it went cleanly and built okay. So that's what I landed on. Maybe it
would be better to use a patch instead of pushing my own fork like this, but
whatever this works just fine. I can always switch to the patch method later.
https://git.zx2c4.com/cgit/commit/?h=jd/zx2c4-deployment&id=8ed1bef90f631989c0cadc326a163b874a64e02d
|
|
I couldn't use llm-ollama because it required some package upgrades, so I
started going down that rabbit hole and ended up 1) realizing that these
packages are way out of date now, and 2) fiddling with overrides to get
everything to work. I finally figured it out, the `postPatch` in ollama-python
was throwing me off for like half a day.
Anyway, one thing to note is that these are changing fast and I need to either
move onto nixpkgs unstable for python stuff, or maintain my own builds of all of
these. Not sure which is more appropriate right now.
Oh and I had to fixup some logging stuff in Biz/Storybook.py because ruff
started complaining about something, which is weird because I don't think the
version changed? But it was easy enough to change.
|
|
OBS wouldn't allow me to stream because there was something wrong with `nvenc`,
which I guess is the nvidia encoding library? Anyway searching around I found
this nvidia-patch overlay and it fixed it right away, so that's cool.
|
|
I realized I don't need this stupid `__main__` convention anymore because my
build system always calls Python programs like `python -m main`, so I just need
to have a function named `main()`. I also started adding some general coding
conventions to the README and fixed a typo.
|
|
The problem was that the ssl cert wasn't renewing. I added the domains to the
nginx config, and cleaned up some other stuff. During debug I also udpated
nixos-mailserver and nixos stable because I thought it would fix something; it
didn't but I might as well use the updated versions of stuff since it seems to
still work fine.
|
|
I mostly wanted a formatter that would format `inherit` blocks
vertically, because otherwise they are super hard to read when diffing
or even just editing. Both alejandra and the new nixos/nixfmt format
verically like this, but alejandra has slightly better format (I guess)
and for some reason nixfmt did not respect my `GLOBIGNORE` setting when
doing `nixfmt **/*.nix` so it was trying to format stuff in `_/nix`, and
failed. So anyway I went with alejandra.
- https://github.com/kamadorueda/alejandra
- https://discourse.nixos.org/t/enforcing-nix-formatting-in-nixpkgs/49506
|
|
It's good to update, I think nixos 24.05 was deprecated anyway. The iPython
dontCheck is because there was a test timeout; I don't even know why iPython is
being pulled in but whatever.
|
|
With run.sh, we can build and run the file in one go. This means we can also use
it as an interpreter in a shebang line and properly use the Unix executable bit.
This is pretty cool and gives a few advantages: running any executable file is
just `exec file.hs` or even `./file.hs`, finding all executables is `fd -t x`,
you don't need to specify or know an `out` name to run something, execution of a
program is standardized.
There is a hack to get this to work. In C and Common Lisp, `#!` is illegal
syntax, so I had to use shell syntax to invoke run.sh, call it on the current
file, and then exit the shell script. Meanwhile, run.sh takes the file and evals
the whole thing, building and running it. As long as either `//` or `;` is a
comment character in the target language, then this works. Maybe a better thing
to do would be to pre-process the file and remove the `#!` before passing it to
the C compiler, like [ryanmjacobs/c][1] and [tcc][2]? However this won't work in
Lisp because then I can't just load the file directly into the repl, so maybe
the comment hack needs to stay.
[1]: https://github.com/ryanmjacobs/c/tree/master
[2]: https://repo.or.cz/tinycc.git/blob/HEAD:/tccrun.c
|
|
This adds the Images endpoint and related functions for loading and saving
images to the filesystem.
In the view layer, it also loads the images asynchronously using HTMX, so the
images get lazy-loaded only when they are done generating.
|
|
This is basically a full rewrite. I ripped out Flask and rearchitected the whole
thing to use fully RESTful resources and endpoints using Ludic. The UI was
completely redone to use Ludic's components. I added tests for everything that I
reasonably could.
This is almost ready for an alpha launch. Before shipping it I still need to:
1. generate images using image n-1 applied to `openai.images.create_variation()`
2. write a nix service, get it on a VM somewhere, I'll probably provision a new
VM for this
3. replace the `db` thing with a real sqlite database
I only need the first one done to show it to Lia and see if she likes it, that
should be completed in a day or two. Then the nix service and deployment won't
take long at all. Setting up a sqlite database will be annoying, but that I
can't see that actually taking more than 2 days. So max 5 days out from
launching this to friends and family.
|
|
This required upgrading to python 3.12 because of some f-string format thing
that ludic uses. It's kind of annoying but the upgrade was easy enough, so I
just did it.
|
|
It's good to do this often.
|
|
This paritally used gptme to create a storybook generator. The problem I ran
into is that gptme doesn't do any architecting or considerations for
maintainable code, or even readable code, so it just wrote a long script. I
couldn't test it. Also, it didn't actually generate a 10-page story, it
generated 10 separate stories. So, I ended up writing it myself and using gptme
to fixup TODOs that I wrote along the way.
|
|
I was getting confused about what is a product and what is internal
infrastructure; I think it is good to keep those things separate. So I moved a
bunch of stuff to an Omni namespace, actually most stuff went there. Only things
that are explicitly external products are still in the Biz namespace.
|
