Bring back commit 3844, albeit in simplified form. I'd forgotten that the
one place where we still need to buffer rendering is when people hold down
up/down arrow keys.
Bugfix of commit 3850 for the sandbox/ app. I'd hoped to just quickly move
past this ugly approach, but a cleaner way is more involved than I thought.
This way is ugly partly because I'm introducing a bunch of conditionals
without testing them. One or more of my additions may well be hiding bugs.
Or I may need to add them in a few other places.
The clean way is to update the fake screen model to accurately mimic the
new real screen, where out of bounds prints aren't silently ignored, and
where scrolling is a fact of life.
Now it's much more apparent why things are slow. You can see each repaint
happening. Already I fixed one performance bug -- in clear-rest-of-screen.
Since this subverts Mu's fake screen there may be bugs.
Another salubrious side effect: I've finally internalized that switching
to raw mode doesn't have to clear the screen. That was just an artifact
of how termbox abstracted operations. Now I can conceive of using termbox
to build a repl as well.
(I was inspired to poke into termbox internals by
http://viewsourcecode.org/snaptoken/kilo and
https://github.com/antirez/linenoise)
Standardize the order of some common blocks in `render`, `render-text`
and `render-code`. This is preparation for trying to reorganize them to
reduce duplicate code.
I accidentally got rid of git snapshotting of lessons back when I switched
to testable file primitives last December (commit 3705).
>:-(
Bringing it back now, hopefully better. The improvement is that there's
now at most one commit every time we hit F4.
This change adds yet another reason that running `mu` from a different
directory is just not supported.
Reorder products of some functions in the edit/ and sandbox/ apps. My
recent realization: always return 'real' products before ones that just
indicate an ingredient is mutable.
Update sandbox/ with recent changes to edit/ (commit 3695 onwards).
[Incidentally, this is the first commit to be made while running on
OpenBSD. Simulated and host systems are going to blur together from now
on.]
Periodic cleanup to replace 'reply' with 'return' everywhere in the
repo.
I use 'reply' for students to help reinforce the metaphor of function
calls as being like messages through a pipe. But that causes 'reply' to
get into my muscle memory when writing Mu code for myself, and I worry
that that makes Mu seem unnecessarily alien to anybody reading on
Github.
Perhaps I should just give it up? I'll try using 'return' with my next
student.
Cleaning up the console interfaces before we start changing the socket
interfaces to look like them. Reading from sockets need to be
non-blocking just like reading from the console.
Stop requiring jump instructions to explicitly provide a ':label' type
for jump targets.
This has been a source of repeated confusion for my students:
a) They'd add the ':label' to the label definition rather than the
jump target (label use)
b) They'd spend time thinking about whether the initial '+' prefix was
part of the label name.
In the process I cleaned up a couple of things:
- the space of names is more cleanly partitioned into labels and
non-labels (clarifying that '_' and '-' are non-label prefixes)
- you can't use label names as regular variables anymore
- you can infer the type of a label just from its name
Redo commit 3457.
Basically there were 3 unicode characters we changed back then:
solid horizontal line: 9473 -> 9472
fuzzy horizontal line: 9480 -> 9548
fuzzy vertical line: 9482 -> 9550
The solid horizontal line has no issues, so we just redo it here.
For the other two, we'll perform the substitution only when rendering
html. That gives us the best of both worlds: the scenario screens render
right in html, and alt-tabbing continues to be snappy when running the
edit/ app.
Revert commit 3457, where I switched the unicode characters used in the
edit/ app to something that doesn't render double-wide in html. It turns
out that the new unicode characters made iTerm2 sluggish in alt-tabbing
between windows. (Commit 3488 only fixed the screen-clearing issue.)
I haven't reverted the html files. I'm going to redo commit 3457 next so
the html files continue to render like they do now.
Ugly that we didn't need 'screen' to provide a type in scenarios
(because assume-screen expands to a definition of 'screen') but we did
need a type for 'console'. Just never require types for special names in
scenarios.
A long-standing problem has been that I couldn't spread code across
'run' blocks because they were separate scopes, so I've ended up making
them effectively comments. Running code inside a 'run' block is
identical in every way to simply running the code directly. The 'run'
block is merely a visual aid to separate setup from the component under
test.
In the process I've also standardized all Mu scenarios to always run in
a local scope, and only use (raw) numeric addresses for values they want
to check later.
Well, almost. I can't use them in some places in C++ where I'm just
creating a temporary reagent without passing it through transforms. Like
in some unit tests. I can't use them in memory-should-contain.
And there's one remaining bug: I can't use abbreviations in a couple of
places in 075channel.mu.
Done using 'text' type abbreviation everywhere.
There's still a problem. If we define a function with a type
abbreviation and then redefine it without, I think we end up creating
separate variants. That seems wrong. Let's isolate a scenario for that
next.
Thanks Sam Putman for helping think through this idea.
When you encounter a backslash, strip it out and pass through any
following run of backslashes. If we 'escaped' a single following
character like C, then the character '\' would be the same as:
'\\' escaped once
'\\\\' escaped twice
'\\\\\\\\' escaped thrice (8 backslashes)
..and so on, the number of backslashes doubling each time. Instead, our
approach is to make the character '\' the same as:
'\\' escaped once
'\\\' escaped twice
'\\\\' escaped thrice
..and so on, the number of backslashes merely increasing by one each
time.
This approach only works as long as backslashes aren't also overloaded
to create special characters. So Mu doesn't follow C's approach of
overloading backslashes both to escape quote characters and also as a
notation for unprintable characters like '\n'.
This is part of efforts to allow students to transition gradually from
the sandbox to running programs directly on the commandline, writing
real scenarios, etc. Running on the commandline requires 'main', but
overriding 'main' would mess up edit/ which is itself a Mu program.
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.
In some rare situations the editor would join a line with the next when
it should simply wrap to the next screen row. Thanks Caleb and Ella
Couch for finally running into a situation that was easy to reproduce.
The scenario diffs are misleading on this commit. I had to:
a) delete the obsolete 'editor-wraps-cursor-after-inserting-characters'
because it was written back when a line just large enough to fit in a
single line would not wrap:
| | <-- screen boundary
abcde
These days it will wrap after making room for the wrap indicator:
| | <-- screen boundary
abcd↩
e
b) rename editor-wraps-cursor-after-inserting-characters-2 to
editor-wraps-cursor-after-inserting-characters-in-middle-of-line
c) create a new scenario demonstrating the bug:
editor-wraps-cursor-after-inserting-characters-at-end-of-line
Standardize quotes around reagents in error messages.
I'm still sure there's issues. For example, the messages when
type-checking 'copy'. I'm not putting quotes around them because in
layer 60 I end up creating dilated reagents, and then it's a bit much to
have quotes and (two kinds of) brackets. But I'm sure I'm doing that
somewhere..
In the process I've also simplified the sandbox/ app. Since it's
impossible for sandbox editors to span multiple pages, we can drop all
scroll support altogether.
This brings back some of the complexity I thought I'd gotten rid of in
2799.
The regression brought home the point that I'd forgotten to write tests
for those bits. Written now.
It also brought home the point that our cleanup in 'reload' has always
been hacky and incomplete.
It's also ugly that those tests in the sandbox/ and edit/ apps need
changing (particularly when the test is about how the output doesn't
change).
I'm dropping all mention of 'recipe' terminology from the Readme. That
way I hope to avoid further bike-shedding discussions while I very
slowly decide on the right terminology with my students.
I could be smarter in my error messages and use 'recipe' when code uses
it and 'function' otherwise. But what about other words like ingredient?
It would all add complexity that I'm not yet sure is worthwhile. But I
do want separate experiences for veteran programmers reading about Mu on
github and for people learning programming using Mu.
Thanks Nicolas Léveillé for running up against this bug:
https://news.ycombinator.com/item?id=11094837
(Also noticed and fixed several subsidiary issues. This whole aspect
doesn't seem fully baked yet.)
If you restore 2 sandboxes, the first render was setting
response-starting-row-on-screen on both, without ever rendering a
response. If the lower sandbox contained a print and rendered the screen
instead of the response, the original response-starting-row-on-screen
was never reset. If the process of running the sandboxes caused the
lower sandbox's title bar to move below the now-stale
response-starting-row-on-screen[1], editing the lower sandbox no longer
works.
[1] (Either because the upper sandbox prints to screen as well (causing
the first F4 to move the lower sandbox down by several lines), or
because a fresh sandbox is created with several lines before the first
F4 is hit.)
Current solution: never set response-starting-row-on-screen during
reload (or otherwise when there's no response).
This is hard to test right now because 'restore' is not a tested
interface, and I can't come up with another situation where the
response-starting-row-on-screen goes stale. So I'm now trying to keep
all changes to response-starting-row-on-screen close together. Another
idea is to add a check that the click row lies below the
response-starting row *and* above the start of the next sandbox. (But
what if there's no next sandbox?)
(This bug is really a regression, introduced last Sep in 2163.)
I'd feared that the refcount errors in the previous commit meant there
was a bug in my ref-counting, so I temporarily used new variables rather
than reusing existing ones. But it turns out the one remaining place
memory corruption can happen is when recipes don't use default-scope and
so end up sharing memory. Don't I have a warning for this?
When I first forked it from the edit/ app, I wasn't sure how to deal
with changing the recipe side when the only way the program accesses it
is with the untestable 'restore' hack. Now we introduce a little hook
into event-loop and pass in any updated recipe side directly.
In the process I've cleaned up several minor stylistic inconsistencies
between edit/ and sandbox/ apps.
I'm now thinking about how to support scrolling on the sandbox side.
Caleb's idea is to use down-arrow inside the sandbox editor, and then
"scroll off" the editor to the top of each successive sandbox. I think
I'll reserve the white background as the cursor color in that situation.
I wonder if I should just undo all the support for sandbox labels since
yesterday. Labels are perhaps superfluous once I support scrolling and
reorder sandboxes to always throw ones with errors up top. But then you
can end up scrolling through lots of tests without any sense of how far
down you are. So the other approach is to keep labels and try to keep
them stable, not reorder them.
Looking further ahead I'm going to need a way to jump to a specific
sandbox. Maybe instead of reordering sandboxes I should just
automatically render from the first sandbox with error. Maybe show the
number of failed sandboxes in the status instead of the index of the
first failure.
It also seems useful that the number maps to the name of the file the
sandbox is saved in. However this mapping is currently a happy accident
and not actually tested.
I'm starting to switch gears and help make the editor useable with
many many sandboxes. This is just the first step of several.
This is the one major refinement on the C programming model I'm planning
to introduce in mu. Instead of Rust's menagerie of pointer types and
static checking, I want to introduce just one new type, and use it to
perform ref-counting at runtime.
So far all we're doing is updating new's interface. The actual
ref-counting implementation is next.
One implication: I might sometimes need duplicate implementations for a
recipe with allocated vs vanilla addresses of the same type. So far it
seems I can get away with just always passing in allocated addresses;
the situations when you want to pass an unallocated address to a recipe
should be few and far between.
Still can't print non-integer numbers, so this is a bit hacky.
The big consequence is that you can't print literal characters anymore
because of our rules about how we pick which variant to statically
dispatch to. You have to save to a character variable first.
Maybe I can add an annotation to literals..
