search

w3c / aria-at-automation-driver

A WebSocket server which allows clients to observe the text enunciated by a screen reader and to simulate user input
Other
10 stars 6 forks source link

Improve platform support and requirements documentation #15

Closed jugglinmike closed 1 year ago

jugglinmike commented 2 years ago

We've been trying to limit this project's runtime dependencies. Previously, we ensured that running the driver does not require the Visual Studio development suite (despite our use of it in development) by testing with a Microsoft-provided "development environment" virtual machine for Windows.

Today, I attempted to run the driver in a virtual machine created from a generic Windows installation disc image. This surfaced a couple problems which were not apparent before:

The commits in this pull request address both concerns

mzgoddard commented 1 year ago

I want to split this PR so we can merge part of it.

The README documentation on the VC++ runtime I think would be good to merge. I'm not sure how to easily confirm its needed. But it seems a reasonable thing to note. It is pointing the right version that was released alongside Visual Studio 2019.

I think including robotjs's build in the repo is a quick solution to the problems it solves. I am conflicted about accepting it. Publishing it as part of the npm package while odd, it isn't a part of the repo's public api. Including it in the repo opposes the common advice to not store large binary files in a git repo. I've been trying to test the version in the PR but I'm blocked with the demo. I haven't been able to recently get NVDA to emit speech events.

One option I want to compare the in-repo robotjs node module with is depending on a release of robotjs on github by the user who posted the PR mentioned in the README change of this PR.

I want to think on this more but wanted to post something here.

jugglinmike commented 1 year ago

You got it!