gfngfn/SATySFi#192

It would be useful to have a class file such that

• it takes a single block-boxes
• and creates a document with single page where page size fits the given block-boxes

We want to use such a class file when we create a figure with SATySFi which will be embedded in another document ofLaTeX or Word.

https://qiita.com/wasabiz/items/7351e9ccd55fd58ba888

According to the document,

%% If [fill-tail] is true, [Inline.fil] is added at the end of the return value.

If fill-tail is false, there may be no Inline.fil in the inline text and sometimes line-break fails to break it into lines. It seems more user-friendly to set fill-tail as true by default because that's what we expect in most of the cases.

Please send a message saying "let me join!" in this thread if you wish to become a contributor of this project. Note that, however, being a contributor of this repository does not mean you are fully able to push arbitrary changes to master. Usually we follow the following workflow to make changes in the code in master:

• create a pull request
• get it reviewed from a contributor who knows well the code base you are changing
• once your reviewer approved your pull request, merge it to master

That is, even a non contributor can send us pull requests and even a contributor has to create a pull request whenever, so becoming a contributor is not very important if you just want to send us changes. Instead contributors are assigned reviewing tasks and responsible for checking the given patch works fine.

Background

At present, we have unit tests and regression tests to keep quality. We write unit tests by writing all test results to output and compare them with snapshots because we don't have unit testing frameworks for SATySFi.

I published the very first version of satysfi-test. I consider we can try it out, and replace all unit tests to ones with satysfi-test if it works well.

Goal

We should have better unit tests using satysfi-test.

@zeptometer Can I assign you this issue?

We want more precise implementations of Float.ceil, Float.floor, etc than ZP0.ceil, ZP0.floor, etc.

Background

Currently, we use jest-image-snapshot for regression. Tests compile test SATySFi files to pdf files and then convert to images, and finally compare them with snapshots. This was kind of a compromise plan, give that we don't have very good tooling for comparing pdf snapshot directly.

That's why I developed jest-pdf-snapshot. Although it is still unstable, I believe this will helps us have better regression tests.

Goal

Replace the current regressions tests with jest-image-snapshot with those with jest-pdf-snapshot.

Derive provides typesetting commands for displaying proof trees. We included Drive in satysfi-base for convenience when we didn't have a package management system.

Now that Satyrographos is widely used in the SATySFi community, it's better to have it as an independent package from satysfi-base. Derive provides a different feature from the core concern of satysfi-base.

the date when I am assigned to the SATySFi advent calendar 2019.

• rename satysfi-lib to satysfi-base
• update README
• add a package file for satyrographos
• add LICENSE
• publish on satyrographos-repo

gfngfn/SATySFi#202
We got a way to compare or hash strings!

We may want to add tests to confirm that we don't break typesetting libraries (like derive, code2, ... etc).

My current idea is leveraging existing tools for front-end development as a first step:

• jest-image-snapshot
• reg-suit

typeset/code2.satyh uses DejaVuSansMono. The trouble is that the vanilla SATySFi doesn't provide the font and users need to install it manually. We can avoid this by adding SATySFi-fonts-dejavu as satyrographos dependency.

Since version 0.0.5 https://github.com/gfngfn/SATySFi/blob/master/CHANGELOG.md#005---2020-07-11 SATySFi provides primitives log and exp for calculation of logarithm and exponentiation.

Aside from #55 and #96, I would appreciate if you could release a new minor version instantly when every new feature has been added unless it's considered conflicting with existing part. How do you think @nyuichi ?

I have not tested thoroughly, but it seems that they do not behave correctly. For example, List.drop 0 [3; 1; 4] evaluates to [1; 4].

RegExp.exec is very slow for the following example.

@require: stdja
@require: base/regexp

let longString = ` I do not know why but it seems code2 is slow if it contains a long line.
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo
wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo wakegawakaranaiyo` in 
%% This is slow
let slowRegex = RegExp.seq RegExp.bof RegExp.space in
%% These are not slow
let notSlowRegexList = [
    RegExp.bof;
    RegExp.space;
    RegExp.seq RegExp.bof RegExp.alpha;
    ] in
let _ = RegExp.exec slowRegex longString in
let _ = List.map (fun notSlowRegex -> RegExp.exec notSlowRegex longString) notSlowRegexList in
document (|
  title = {slow regexp};
  author = {MasWag};
  show-title = false;
  show-toc = false;
|) '<>

In my environment, a single call of RegExp.exec + (almost minimal typesetting) took about 30 seconds.

real	0m33.527s
user	0m33.334s
sys	0m0.040s

Currently, this repo has two dependencies on JS(node/yarn)

• unit tests (being replaced with #142)
• scripts

Considering their verboseness, we want to get rid of JS dependency and find an alternative that is more concise.

gfngfn/SATySFi#210

Changelog

• #278 Add some string functions by puripuri2100
• #280 Fix 279 by puripuri2100
• #198 Add big-int by puripuri2100
• #286 Support SATySFi 0.0.x

TODOs

For detailed steps, refer to 素敵なライブラリをSatyrographosで配布しよう！.

• Update opam and Satyristes → #282
• Crete tag 1.6.0
• Create PR to satyrographos repo → na4zagin3/satyrographos-repo#653
• Get the PR merged

https://twitter.com/damedemoyarou/status/1214781637306667008

With satysfi-base v1.2.1 on Satyrographos, we aren't able to @require: base/bool, are we?

The current version of SATySFi is 0.0.7 (NOTE: 0.0.8 is not available at opam?), while the current release of satysfi-base supports 0.0.6. satysfi-base should support the latest SATySFi.

Refer to:

• Changelog 0.0.6->0.0.7
• Changelog 0.0.7->0.0.8

To be clarified:

• Which version to support?
  • likely 0.0.5-0.0.7
  • How do we ensure that it works as expected for each version?

TODO

• update dependency in opam
• refine CI to run tests against different versions of satysfi

I found the Block module in block.satyh has the following definition

  let frame-breakable pads deco f ctx =
    block-frame-breakable ctx pads deco f

while it doesn't have frame-breakable in its signature.

TODOs

For detailed steps, refer to 素敵なライブラリをSatyrographosで配布しよう！.

• Update opam and Satyristes → #257
• Crete tag 1.5.0
• Create PR to satyrographos repo
• Get the PR merged