Towards the prim module by nomeata · Pull Request #1088 · caffeinelabs/motoko

nomeata · 2020-01-09T16:50:07Z

this starts implementing the plan in
#867 (comment)

No values are in scope by default.
In bare uses of Motoko: Use import Prim "mo:prim" to get
Prim.debugPrint etc.
In real uses, the stdlib should provide everything you need.

Quite some churn in the test suite. Went through run and fail so
far.

this starts implementing the plan in #867 (comment) No values are in scope by default. In bare uses of Motoko: Use `import Prim "mo:prim"` to get `Prim.debugPrint` etc. In real uses, the `stdlib` should provide everything you need. Quite some churn in the test suite. Went through `run` and `fail` so far.

nomeata · 2020-01-09T17:21:09Z

I hit a snag: Nested modules.

Although they are not supported in drun, we do use local actors in our test suite, e.g. in:

import Prim "mo:prim";
Prim.debugPrint (debug_show true);
let _ = actor {
  Prim.debugPrint (debug_show false);
  Prim.debugPrint (debug_show 1);
};
()

Previously this compiled due to a horrible hack about prelude definitions in the backend. But the hack does not apply to Prim (which looks like any old let-binding then), so this is now a non-closed actor.

What to do? Maybe simply stopping to use local actors in the test suite (or only in the interpreter)?

I guess there are actually just a few of these. Disabling on drun/stub seems to be fine.

(Working on updating test/drun)

and move the test projects into its own directory, to avoid spurious recompilation of the `lsp-int` binary.

@ggreif

but somehow this does not compile. @ggreif, can you help?

nomeata · 2020-01-09T18:30:55Z

@ggreif I have some problem adjusting test/random. I would expect joachim/prim-module-qc to work. Maybe the msum is doing something strange. Would you mind having a look at that commit and help me out (e.g. by pushing one that works :-)?

ggreif · 2020-01-10T10:50:23Z

@ggreif I have some problem adjusting test/random. I would expect joachim/prim-module-qc to work. Maybe the msum is doing something strange. Would you mind having a look at that commit and help me out (e.g. by pushing one that works :-)?

I am working on this. Just for the record, one can run a subset of the tests by (e.g.)

env PATH=$PATH:src $(nix-build -A qc-motoko)/bin/qc-motoko -p failures

which will leave behind specific .mo files, which then can be inspected/massaged in isolation.

It appears that "import Prim ..." is prefixed on every line, but it is only allowed at the top of the program. This is for the successes test (corresponding to "tests.mo"), all the others are one-liners (IIRC), so they should be fine.

ggreif · 2020-01-10T12:07:53Z

@nomeata please merge origin/joachim/prim-module-qc, I'll watch the CI.

QC fixes for #1088

nomeata · 2020-01-10T14:06:43Z

Ok, CI is happy.

…rim-module

kritzcreek

This looks great, do you think we want to hold off on this until after Davos? This is definitely going to break code in linkedup (having to add the Prim import) and I don't know when the next DFX release is happening so we might end up dropping this change on them very shortly before the demo.

kritzcreek · 2020-01-10T14:12:31Z

src/pipeline/url.ml

   parse "mo:foo/bar"     = Ok (Package ("foo", "bar"))
   parse "mo:foo"         = Ok (Package ("foo", ""))

+   parse "mo:prim"        = Ok (Prim)


Not saying it needs to happen in this PR, but these comments look like they should be unit tests?

kritzcreek · 2020-01-10T14:13:33Z

src/prelude/prelude.ml

+This module should contain everything that cannot be implemented in plain
+Motoko. It is available via `import Prim "mo:prim"`. Normal user code would
+usually not import that module directly, but through the stdlib, which takes
+care of providing a prober module structure, e.g. exposing Array_tabulate


Suggested change

care of providing a prober module structure, e.g. exposing Array_tabulate

care of providing a proper module structure, e.g. exposing Array_tabulate

ggreif

This is big. I have just skimmed the surface. But I agree, this needs to be done before lots of big (d)apps grow in the wild. Is LinkedUp informed about upcoming changes? Also @lsgunnlsgunn should be in the loop so she is prepared.

design/DFX-Interface.md

src/ir_def/check_ir.ml

src/pipeline/pipeline.ml

Co-Authored-By: Gabor Greif <gabor@dfinity.org>

nomeata · 2020-01-10T14:37:58Z

This is big.

Actually it isn’t so big:

In prelude.ml you can see the prelude and the prim.
A little bit of code in the pipeline to make "mo:prim" work.
A lot of mechanical updating of the testing.

Would you prefer one PR that adds the infrastructure (unchanged prelude, empty prim_module), and a separate that actually moves things over?

Is LinkedUp informed about upcoming changes? Also @lsgunnlsgunn should be in the loop so she is prepared.

Judging from updating the produce exchange, there isn’t much to do. Outside users should usually not import Prim, but should use the stdlib (e.g. Array.tabulate, Iter.range).

The biggest one is Mostly debugPrint, which should be properly exported through the stdlib (maybe Debug.print?). We can do that in a separate PR shortly afterwards.

We can maybe port LinkedUp ourself and provide a PR with the changes needed for that.

nomeata · 2020-01-10T14:39:56Z

Would you prefer one PR that adds the infrastructure (unchanged prelude, empty prim_module), and a separate that actually moves things over?

Actually, that sounds like a good idea. Will do that.

this make `import "mo:prim"` import the module in the `prim_module` string in `prelude.ml`. This is currently empty; the code is added in PR #1088.

nomeata · 2020-01-10T14:46:46Z

See #1093. Now with a warning upon --package prim ..

* Infrastructure for the hard-coded prim module this make `import "mo:prim"` import the module in the `prim_module` string in `prelude.ml`. This is currently empty; the code is added in PR #1088. * Print an error upon `--package prim ...` * Update src/prelude/prelude.ml Co-Authored-By: Gabor Greif <gabor@dfinity.org> Co-authored-by: Gabor Greif <ggreif@gmail.com>

…rim-module

test/perf/reversi.mo

test/run-drun/create-then-trap.mo

test/run/hashes.mo

ggreif

I think this is good to go, with following caveats:

heads-up to @enzoh and other teams
needs mention in Motoko user manual
snippets for range and hashInt so that users can include it themselves.

Please choose the merge time with consideration of the Bronze situation.

nomeata · 2020-01-10T16:01:16Z

This looks great, do you think we want to hold off on this until after Davos?

The merge conflicts in the test suite will not be fun, but maybe that’s reasonable.

OTOH, if we prepare the stdlib so that they don’t need to import prim (just stdlib/Debug), and we give them a PR against LinkedUp with the changes, then it would be fine, wouldn't it?

Co-Authored-By: Gabor Greif <gabor@dfinity.org>

this anticipates #1088: With this change, the LinkedUp code can be changed to use only the stdlib, but _no_ definitions from the prelude. This means that #1088 would not break LinkedUp.

nomeata · 2020-01-10T16:56:56Z

See dfinity/linkedup#3 for how small the fallout on Linkedup is

kritzcreek · 2020-01-10T17:01:36Z

If they're fine with it that's great. I just didn't want to put one more thing on their plate.

* Stdlib: Export some conversion functions also in Nat and Int this anticipates #1088: With this change, the LinkedUp code can be changed to use only the stdlib, but _no_ definitions from the prelude. This means that #1088 would not break LinkedUp. * Likewise, export Array.{init/tabulate} * Expose Debug.print

…rim-module

nomeata · 2020-01-10T17:14:17Z

Yes, that’s absolutely right in asking for caution here (and I am glad you made me look at LinkedUp).

nomeata force-pushed the joachim/prim-module branch from 3a55d9f to ee0d31d Compare January 9, 2020 16:51

Use Prim module in test/run-drun where needed

8bc9992

nomeata force-pushed the joachim/prim-module branch from f606155 to 8bc9992 Compare January 9, 2020 17:25

Update test/perf

5cd22b9

nomeata mentioned this pull request Jan 9, 2020

Clean up prelude #867

Closed

nomeata added 6 commits January 9, 2020 18:41

Fix test.lsp

8a423cd

and move the test projects into its own directory, to avoid spurious recompilation of the `lsp-int` binary.

Odd (but good?) effects on tests/mo-idl

0da9822

Update tests/trap

496a233

Update stdlib

741c651

Fix tests/repl

b7a4c47

Try to fix tests.random

15e3f1c

but somehow this does not compile. @ggreif, can you help?

Fix tests/repl

c3b6a88

nomeata force-pushed the joachim/prim-module branch from b7a4c47 to c3b6a88 Compare January 9, 2020 18:32

Update samples

7e6f2ee

ggreif added 2 commits January 10, 2020 11:55

fix test files' extension

60df728

prepend import Prim to file, not all lines

e836fa7

nomeata added 2 commits January 10, 2020 14:58

Merge pull request #1091 from dfinity-lab/joachim/prim-module-qc

2725f9e

QC fixes for #1088

Update repl output

435ca9f

nomeata mentioned this pull request Jan 10, 2020

Completions for Prim #1092

Merged

nomeata marked this pull request as ready for review January 10, 2020 14:06

Merge branch 'master' of github.com:dfinity-lab/motoko into joachim/p…

701265d

…rim-module

kritzcreek reviewed Jan 10, 2020

View reviewed changes

nomeata added 2 commits January 10, 2020 15:22

typo

1f3bbac

Update stdlib/TrieMap

0f736b2

ggreif reviewed Jan 10, 2020

View reviewed changes

design/DFX-Interface.md Show resolved Hide resolved

src/ir_def/check_ir.ml Outdated Show resolved Hide resolved

src/pipeline/pipeline.ml Outdated Show resolved Hide resolved

nomeata and others added 3 commits January 10, 2020 15:33

Quick fix to avoid printing all of prim in repl

899b099

Update src/pipeline/pipeline.ml

4c560e0

Co-Authored-By: Gabor Greif <gabor@dfinity.org>

Update src/ir_def/check_ir.ml

4df839c

Co-Authored-By: Gabor Greif <gabor@dfinity.org>

nomeata added a commit that referenced this pull request Jan 10, 2020

Infrastructure for the hard-coded prim module

a0b180f

this make `import "mo:prim"` import the module in the `prim_module` string in `prelude.ml`. This is currently empty; the code is added in PR #1088.

nomeata mentioned this pull request Jan 10, 2020

Infrastructure for the hard-coded prim module #1093

Merged

More updating of test output

363391f

Merge branch 'master' of github.com:dfinity-lab/motoko into joachim/p…

8a8cc93

…rim-module

ggreif reviewed Jan 10, 2020

View reviewed changes

test/perf/reversi.mo Show resolved Hide resolved

ggreif reviewed Jan 10, 2020

View reviewed changes

test/run-drun/create-then-trap.mo Outdated Show resolved Hide resolved

ggreif reviewed Jan 10, 2020

View reviewed changes

test/run/hashes.mo Show resolved Hide resolved

ggreif approved these changes Jan 10, 2020

View reviewed changes

Update test/run-drun/create-then-trap.mo

e3ad2da

Co-Authored-By: Gabor Greif <gabor@dfinity.org>

nomeata mentioned this pull request Jan 10, 2020

Stdlib: Export more prelude stuff via the stdlib #1096

Merged

nomeata added 2 commits January 10, 2020 18:10

Merge branch 'master' of github.com:dfinity-lab/motoko into joachim/p…

a7ede2c

…rim-module

stdlib: Add Blob.hash(), while I am at it

40013c5

nomeata merged commit 8e485d3 into master Jan 10, 2020

nomeata deleted the joachim/prim-module branch January 10, 2020 17:34

	care of providing a prober module structure, e.g. exposing Array_tabulate
	care of providing a proper module structure, e.g. exposing Array_tabulate

Conversation

nomeata commented Jan 9, 2020

Uh oh!

nomeata commented Jan 9, 2020

Uh oh!

nomeata commented Jan 9, 2020

Uh oh!

ggreif commented Jan 10, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ggreif commented Jan 10, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

nomeata commented Jan 10, 2020

Uh oh!

kritzcreek left a comment

Choose a reason for hiding this comment

Uh oh!

kritzcreek Jan 10, 2020

Choose a reason for hiding this comment

Uh oh!

kritzcreek Jan 10, 2020

Choose a reason for hiding this comment

Uh oh!

ggreif left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

nomeata commented Jan 10, 2020

Uh oh!

nomeata commented Jan 10, 2020

Uh oh!

nomeata commented Jan 10, 2020

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ggreif left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

nomeata commented Jan 10, 2020

Uh oh!

nomeata commented Jan 10, 2020

Uh oh!

kritzcreek commented Jan 10, 2020

Uh oh!

nomeata commented Jan 10, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

ggreif commented Jan 10, 2020 •

edited

Loading

ggreif commented Jan 10, 2020 •

edited

Loading

ggreif left a comment •

edited

Loading