Clean up the rat's nest that all my trace management globals had
gradually turned into.
a) Get rid of 'Start_tracing'. Horryibly named, I don't know how I
missed that until now.
b) Never use START_TRACING_UNTIL_END_OF_SCOPE in main(). It's
confusing to combine it with atexit(delete Trace_stream), because the
atexit() never has to run. Instead we'll just manually initialize
Trace_stream and let atexit() clean up.
c) If we run tests we only want a trace for the test run itself. So
delete the Trace_stream that was initialized at the top of main --
once it's clear we had no load-time errors.
d) Clean up horribly "Load Recipes" waypoints, combine them with the better
name, "Mu Prelude".
Putting these together, we have the following manual tests:
- CFLAGS=-g mu x.mu
Should not create last_run.
- CFLAGS=-g mu --trace x.mu
Should create last_run.
Should write it out exactly once.
- CFLAGS=-g mu --trace x.mu # when x.mu has an error
Should create last_run.
Should write it out exactly once.
- CFLAGS=-g mu --trace test copy_literal # C test
Should create last_run.
Should write it out exactly once.
- CFLAGS=-g mu --trace test recipe_with_header # Mu test
Should create last_run.
Should write it out exactly once.
I don't know how to automate these scenarios yet. We need a way to run
our build toolchain atop our stack.
So far we only checked if a single recipe used a variable with multiple
types in any single space. Now we also ensure that the types deduced for
a variable in a space are identical across recipes.
Undo 3743. Really any time we create new instructions from whole cloth
during rewriting or transform, the whole notion of 'original name' goes
out the window. Pointless trying to fight that fact of life.
One way to ensure we always set old_name is to create a method to
initialize names as opposed to just assigning them.
Still not ideal because we still assign directly most of the time, so
it's easy to forget.
The drawback of this is that we forget to initialize old_name when we
create instructions out of whole cloth in a few places. But this problem
already existed..
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..
More consistent definitions for jump targets and waypoints.
1. A label is a word starting with something other than a letter or
digit or '$'.
2. A waypoint is a label that starts with '<' and ends with '>'. It has
no restrictions. A recipe can define any number of waypoints, and
recipes can have duplicate waypoints.
3. The special labels '{' and '}' can also be duplicated any number of
times in a recipe. The only constraint on them is that they have to
balance in any recipe. Every '{' must be followed by a matching '}'.
4. All other labels are 'jump targets'. You can't have duplicate jump
targets in a recipe; that would make jumps ambiguous.
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.
Thanks Ella Couch for pointing out that Mu was lying when debugging
small numbers.
def main [
local-scope
x:number <- copy 1
{
x <- divide x, 2
$print x, 10/newline
loop # until SIGFPE
}
]
Undo 3272. The trouble with creating a new section for constants is that
there's no good place to order it since constants can be initialized
using globals as well as vice versa. And I don't want to add constraints
disallowing either side.
Instead, a new plan: always declare constants in the Globals section
using 'extern const' rather than just 'const', since otherwise constants
implicitly have internal linkage (http://stackoverflow.com/questions/14894698/why-does-extern-const-int-n-not-work-as-expected)
Move global constants into their own section since we seem to be having
trouble linking in 'extern const' variables when manually cleaving mu.cc
into separate compilation units.
I'd been toying with this idea for some time now given how large the
repo had been growing. The final straw was noticing that people cloning
the repo were having to wait *5 minutes*! That's not good, particularly
for a project with 'tiny' in its description. After purging .traces/
clone time drops to 7 seconds in my tests.
Major issue: some commits refer to .traces/ but don't really change
anything there. That could get confusing :/
Minor issues:
a) I've linked inside commits on GitHub like a half-dozen times online
or over email. Those links are now liable to eventually break. (I seem
to recall GitHub keeps them around as long as they get used at least
once every 60 days, or something like that.)
b) Numbering of commits is messed up because some commits only had
changes to the .traces/ sub-directory.
More reorganization in preparation for implementing recursive abandon().
Refcounts are getting incredibly hairy. I need to juggle containers
containing other containers, and containers *pointing* to other
containers. For a while I considered getting rid of address_element_info
entirely and just going by types for every single
update_refcount. But that's definitely more work, and it's unclear that
things will be cleaner/shorter/simpler. I haven't measured the speedup,
but it seems worth optimizing every pointer copy to make sure we aren't
manipulating types at runtime.
The key insight now is a) to continue to compute information about
nested containers at load time, because that's the common case when
updating refcounts, but b) to compute information about *pointed* values
at run-time, because that's the uncommon case.
As a result, we're going to cheat in the interpreter and use type
information at runtime just for abandon(), just because the
corresponding task when we get to a compiler will be radically
different. It will still be tractable, though.
Kartik Agaram