I've been saying for a while[1][2][3] that adding extra abstractions makes
things harder for newcomers, and adding new notations doubly so. And then
I notice this DSL in my own backyard. Makes me feel like a hypocrite.
[1] https://news.ycombinator.com/item?id=13565743#13570092
[2] https://lobste.rs/s/to8wpr/configuration_files_are_canary_warning
[3] https://lobste.rs/s/mdmcdi/little_languages_by_jon_bentley_1986#c_3miuf2
The implementation of the DSL was also highly hacky:
a) It was happening in the tangle/ tool, but was utterly unrelated to tangling
layers.
b) There were several persnickety constraints on the different kinds of
lines and the specific order they were expected in. I kept finding bugs
where the translator would silently do the wrong thing. Or the error messages
sucked, and readers may be stuck looking at the generated code to figure
out what happened. Fixing error messages would require a lot more code,
which is one of my arguments against DSLs in the first place: they may
be easy to implement, but they're hard to design to go with the grain of
the underlying platform. They require lots of iteration. Is that effort
worth prioritizing in this project?
On the other hand, the DSL did make at least some readers' life easier,
the ones who weren't immediately put off by having to learn a strange syntax.
There were fewer quotes to parse, fewer backslash escapes.
Anyway, since there are also people who dislike having to put up with strange
syntaxes, we'll call that consideration a wash and tear this DSL out.
---
This commit was sheer drudgery. Hopefully it won't need to be redone with
a new DSL because I grow sick of backslashes.
I've extracted it into a separate binary, independent of my Mu prototype.
I also cleaned up my tracing layer to be a little nicer. Major improvements:
- Realized that incremental tracing really ought to be the default.
And to minimize printing traces to screen.
- Finally figured out how to combine layers and call stack frames in a
single dimension of depth. The answer: optimize for the experience of
`browse_trace`. Instructions occupy a range of depths based on their call
stack frame, and minor details of an instruction lie one level deeper
in each case.
Other than that, I spent some time adjusting levels everywhere to make
`browse_trace` useful.
I've been working on this slowly over several weeks, but it's too hard
to support 0 as the null value for addresses. I constantly have to add
exceptions for scalar value corresponding to an address type (now
occupying 2 locations). The final straw is the test for 'reload':
x:num <- reload text
'reload' returns an address. But there's no way to know that for
arbitrary instructions.
New plan: let's put this off for a bit and first create support for
literals. Then use 'null' instead of '0' for addresses everywhere. Then
it'll be easy to just change what 'null' means.
The trouble with rewriting 'unused' to '__attribute__(unused)' is that
if we happen to deliberately introduce '__attribute__(unused)' somehow,
say in the standard headers, then it gets expanded twice to '__attribute__(__attribute__(unused))'.
So we switch to a synonym.
Yet another attempt at decomposing incremental edits in some clean way.
The new idea now is that I need to only modify the screen using a
restricted vocabulary of actions:
render-all
render-recipe-side
render-sandbox-side
render-recipe-errors
render-line-from-cursor
render-line-from-start
erase-line-from-cursor
render-character-at-cursor
erase-character-at-cursor
However, decomposing insert-at-cursor is challenging; how to manipulate
cursor-row and cursor-column without also pretending to print to screen?
Do I need to decompose `editor` into multiple containers so that I can
keep cursor-row and cursor-column with screen modifications? Here's what
`editor` looks like after all layers:
container editor [
data:&:duplex-list:char
top-of-screen:&:duplex-list:char
bottom-of-screen:&:duplex-list:char
before-cursor:&:duplex-list:char
left:num
right:num
bottom:num
cursor-row:num
cursor-column:num
indent?:bool
undo:&:list:&:operation
redo:&:list:&:operation
]
It's not obvious that there's a clean way to split all these fields.
Standardize exit paths. Most layers now don't need to know about termbox.
We can't really use `assert` in console-mode apps; it can't just exit because
we want to be able to check assertion failures in tests.
It's always confusing when `break` refers to a `switch` but `continue`
refers to the loop around the `switch`. But we've done ugly things like
this and `goto` for expedience. However, we're starting to run into cases
where we now need to insert code at every `continue` or `continue`-mimicking
`goto` inside the core interpreter loop. Better to make the loop single-entry-single-exit.
Common things to run after every instruction will now happen inside the
`finish_instruction` function rather than at the `finish_instruction` label.
Be more disciplined about tagging 2 different concepts in the codebase:
a) Use the phrase "later layers" to highlight places where a layer
doesn't have the simplest possible self-contained implementation.
b) Use the word "hook" to point out functions that exist purely to
provide waypoints for extension by future layers.
Since both these only make sense in the pre-tangled representation of
the codebase, using '//:' and '#:' comments to get them stripped out of
tangled output.
(Though '#:' comments still make it to tangled output at the moment.
Let's see if we use it enough to be worth supporting. Scenarios are
pretty unreadable in tangled output anyway.)
This was a large commit, and most of it is a follow-up to commit 3309,
undoing what is probably the final ill-considered optimization I added
to s-expressions in Mu: I was always representing (a b c) as (a b . c),
etc. That is now gone.
Why did I need to take it out? The key problem was the error silently
ignored in layer 30. That was causing size_of("(type)") to silently
return garbage rather than loudly complain (assuming 'type' was a simple
type).
But to take it out I had to modify types_strictly_match (layer 21) to
actually strictly match and not just do a prefix match.
In the process of removing the prefix match, I had to make extracting
recipe types from recipe headers more robust. So far it only matched the
first element of each ingredient's type; these matched:
(recipe address:number -> address:number)
(recipe address -> address)
I didn't notice because the dotted notation optimization was actually
representing this as:
(recipe address:number -> address number)
---
One final little thing in this commit: I added an alias for 'assert'
called 'assert_for_now', to indicate that I'm not sure something's
really an invariant, that it might be triggered by (invalid) user
programs, and so require more thought on error handling down the road.
But this may well be an ill-posed distinction. It may be overwhelmingly
uneconomic to continually distinguish between model invariants and error
states for input. I'm starting to grow sympathetic to Google Analytics's
recent approach of just banning assertions altogether. We'll see..
Follow-up to commit 3321: move get_base_type() more thoroughly to layer
55. The notion of a base_type doesn't really make sense before we
introduce type ingredients and shape-shifting containers, and it
simplifies early layers a *lot* even including the cost of that *ugly*
preamble in layer 55 to retrofit all the places.
Don't crash on bad types.
I need to be more careful in distinguishing between the two causes of
constraint violations: bad input and internal bugs. Maybe I should
create a second assert() to indicate "this shouldn't really be an
assert, but I'm too lazy to think about it right now."
Extract a helper to compute the element type of an array. As a side
effect, the hack for disambiguating array:address:number and
array:number:3 is now in just one place.
Programming languages need some higher-level language construct that's
neither an interface nor a class nor an object but a *collection of
mutually recursive functions with a well-defined set of entry points and
common ingredients. Perhaps the solution here is the Haskell "save your
boilerplate" paper. For now I'm going to include the purpose in
auxiliary variable names that aren't really necessary for the core
processing of a function.
Thanks Caleb Couch for reporting this issue.
Tag all transforms as idempotent or not.
I'd fallen off this wagon.
I might even be getting it wrong. Something a type system should
automatically verify.
Always check if next_word() returned an empty string (if it hit eof).
Thanks Rebecca Allard for running into a crash when a .mu file ends with
'{' (without a following newline).
Open question: how to express the constraint that next_word() should
always check if its result is empty? Can *any* type system do that?!
Even the usual constraint that we must use a result isn't iron-clad: you
could save the result in a variable but then ignore it. Unless you go to
Go's extraordinary lengths of considering any dead code an error.
Better implementation of commit 3445: not requiring types for special
variables in scenarios. It turned out that it wasn't working anytime we
needed to call 'get' on a special variable inside a scenario. After
moving that work to an earlier transform we can now use 'filesystem'
without a type inside scenarios.
The old approach with '&' and '@' modifiers turned out to be a bad idea
because it introduces notions of precedence. Worse, it turns out you
want different precedence rules at different times as the old test
alluded:
x:@number:3 # we want this to mean (address number 3)
x:address:@number # we want this to mean (address array number)
Instead we'll give up and focus on a single extensible mechanism that
allows us to say this instead:
x:@:number:3
x:address:@:number
In addition it allows us to shorten other types as well:
x:&:@:num
type board = &:@:&:@:char # for tic-tac-toe
Hmm, that last example reminds me that we don't handle abbreviations
inside type abbreviation definitions so far..
Clean up another case (after commit 3309) of premature support for
shape-shifting recipes, where early layers had code without
corresponding tests.
One addendum to commit 3309: the proximal cause for triggering the
rewrite of type_trees was that I realized to_string() and variants were
lying to me while debugging; they couldn't distinguish between `(a . b)`
and `((a) . b)`
Allow type-trees to be ordered in some consistent fashion. This could be
quite inefficient since we often end up comparing the four sub-trees of
the two arguments in 4 different ways. So far it isn't much of a time
sink.
Rip out everything to fix one failing unit test (commit 3290; type
abbreviations).
This commit does several things at once that I couldn't come up with a
clean way to unpack:
A. It moves to a new representation for type trees without changing
the actual definition of the `type_tree` struct.
B. It adds unit tests for our type metadata precomputation, so that
errors there show up early and in a simpler setting rather than dying
when we try to load Mu code.
C. It fixes a bug, guarding against infinite loops when precomputing
metadata for recursive shape-shifting containers. To do this it uses a
dumb way of comparing type_trees, comparing their string
representations instead. That is likely incredibly inefficient.
Perhaps due to C, this commit has made Mu incredibly slow. Running all
tests for the core and the edit/ app now takes 6.5 minutes rather than
3.5 minutes.
== more notes and details
I've been struggling for the past week now to back out of a bad design
decision, a premature optimization from the early days: storing atoms
directly in the 'value' slot of a cons cell rather than creating a
special 'atom' cons cell and storing it on the 'left' slot. In other
words, if a cons cell looks like this:
o
/ | \
left val right
..then the type_tree (a b c) used to look like this (before this
commit):
o
| \
a o
| \
b o
| \
c null
..rather than like this 'classic' approach to s-expressions which never
mixes val and right (which is what we now have):
o
/ \
o o
| / \
a o o
| / \
b o null
|
c
The old approach made several operations more complicated, most recently
the act of replacing a (possibly atom/leaf) sub-tree with another. That
was the final straw that got me to realize the contortions I was going
through to save a few type_tree nodes (cons cells).
Switching to the new approach was hard partly because I've been using
the old approach for so long and type_tree manipulations had pervaded
everything. Another issue I ran into was the realization that my layers
were not cleanly separated. Key parts of early layers (precomputing type
metadata) existed purely for far later ones (shape-shifting types).
Layers I got repeatedly stuck at:
1. the transform for precomputing type sizes (layer 30)
2. type-checks on merge instructions (layer 31)
3. the transform for precomputing address offsets in types (layer 36)
4. replace operations in supporting shape-shifting recipes (layer 55)
After much thrashing I finally noticed that it wasn't the entirety of
these layers that was giving me trouble, but just the type metadata
precomputation, which had bugs that weren't manifesting until 30 layers
later. Or, worse, when loading .mu files before any tests had had a
chance to run. A common failure mode was running into types at run time
that I hadn't precomputed metadata for at transform time.
Digging into these bugs got me to realize that what I had before wasn't
really very good, but a half-assed heuristic approach that did a whole
lot of extra work precomputing metadata for utterly meaningless types
like `((address number) 3)` which just happened to be part of a larger
type like `(array (address number) 3)`.
So, I redid it all. I switched the representation of types (because the
old representation made unit tests difficult to retrofit) and added unit
tests to the metadata precomputation. I also made layer 30 only do the
minimal metadata precomputation it needs for the concepts introduced
until then. In the process, I also made the precomputation more correct
than before, and added hooks in the right place so that I could augment
the logic when I introduced shape-shifting containers.
== lessons learned
There's several levels of hygiene when it comes to layers:
1. Every layer introduces precisely what it needs and in the simplest
way possible. If I was building an app until just that layer, nothing
would seem over-engineered.
2. Some layers are fore-shadowing features in future layers. Sometimes
this is ok. For example, layer 10 foreshadows containers and arrays and
so on without actually supporting them. That is a net win because it
lets me lay out the core of Mu's data structures out in one place. But
if the fore-shadowing gets too complex things get nasty. Not least
because it can be hard to write unit tests for features before you
provide the plumbing to visualize and manipulate them.
3. A layer is introducing features that are tested only in later layers.
4. A layer is introducing features with tests that are invalidated in
later layers. (This I knew from early on to be an obviously horrendous
idea.)
Summary: avoid Level 2 (foreshadowing layers) as much as possible.
Tolerate it indefinitely for small things where the code stays simple
over time, but become strict again when things start to get more
complex.
Level 3 is mostly a net lose, but sometimes it can be expedient (a real
case of the usually grossly over-applied term "technical debt"), and
it's better than the conventional baseline of no layers and no
scenarios. Just clean it up as soon as possible.
Definitely avoid layer 4 at any time.
== minor lessons
Avoid unit tests for trivial things, write scenarios in context as much as
possible. But within those margins unit tests are fine. Just introduce them
before any scenarios (commit 3297).
Reorganizing layers can be easy. Just merge layers for starters! Punt on
resplitting them in some new way until you've gotten them to work. This is the
wisdom of Refactoring: small steps.
What made it hard was not wanting to merge *everything* between layer 30
and 55. The eventual insight was realizing I just need to move those two
full-strength transforms and nothing else.
Highlight a couple of places where it turns out that we're flying by the
seat of our pants with heuristics, and we don't really understand how to
precompute metadata for a program's types.
Kartik Agaram