Lighter weight standards processes by using a cross-translating PL to craft reference implementations
- Upcoming
- Tentative
- Breakout Sessions
- Upcoming
- Tentative
- Breakout Sessions
Meeting
I'm a PL researcher and I've been designing a programming language to translate well to all the other languages. (Traditional compiler methods based, not AI)
The goal of the tool is to make it easier for different language communities (data scientists/pythonistas, web devs/javascripters, backenders) to identify common problems and share solutions.
With it, you can write a library once and get libraries in all the languages widely used in OSS software.
Some standards efforts have significant components that are not dependent on the stack:
- L10N/I18N standards are about human languages and cultures, not the tech stack that is composing strings to present. Having ubiquitously available libraries like ICU & grapheme clustering would be good.
- Representations of currencies
- How to derive "safe" HTML from untrusted HTML is not about the tech stack.
- Arithmetic about color
- Parsing and unparsing URLs
Perhaps these or similar standards would benefit from define signatures / skeletal types and maybe small test cases in the standard by way of explanation, and in an annex include implementation details and more thorough test suite.
For standards driven mostly by non-software-engineer domain experts, the hope is that a tool like this might allow building reference implementations for many programming language communities with a single engineering point of contact lowering coordination overhead when changes to a standard are needed.
Agenda
Chairs:
Mike Samuel
Description:
I'm a PL researcher and I've been designing a programming language to translate well to all the other languages. (Traditional compiler methods based, not AI)
The goal of the tool is to make it easier for different language communities (data scientists/pythonistas, web devs/javascripters, backenders) to identify common problems and share solutions.
With it, you can write a library once and get libraries in all the languages widely used in OSS software.
Some standards efforts have significant components that are not dependent on the stack:
- L10N/I18N standards are about human languages and cultures, not the tech stack that is composing strings to present. Having ubiquitously available libraries like ICU & grapheme clustering would be good.
- Representations of currencies
- How to derive "safe" HTML from untrusted HTML is not about the tech stack.
- Arithmetic about color
- Parsing and unparsing URLs
Perhaps these or similar standards would benefit from define signatures / skeletal types and maybe small test cases in the standard by way of explanation, and in an annex include implementation details and more thorough test suite.
For standards driven mostly by non-software-engineer domain experts, the hope is that a tool like this might allow building reference implementations for many programming language communities with a single engineering point of contact lowering coordination overhead when changes to a standard are needed.
Goal(s):
Explore how to make the results of standards efforts easier to achieve and available to more developer communities via an emerging technology, and gather requirements from attendees before they require breaking changes in the language
Agenda:
Tentatively:
- Demo the language showing how a well-understood standard could have been developed alongside a reference implementation that cross-compiles to many languages
- Open discussion on language changes that might make the tool fit well into standards workers' workflow
- Open discussion to identify legalese that would need to be addressed before the proposed tool could be relied upon
- Ideally, identify one or more pilot project partners who are willing to trial the project in their work
Materials:
Joining Instructions
Instructions are restricted to W3C users . You need to log in to see them.
Export options
Personal Links
Please log in to export this event with all the information you have access to.
Public Links
The following links do not contain any sensitive information and can be shared publicly.