You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
For beginners, we want distribution, installation, deployment, and invocation to be as simple as possible. Preferably, there would be a very simple tutorial for distribution, installation, and deployment, and a single wrapper script for compiling, testing, and running.
Just as a note, I believe the original vision from Zamua's thesis is that you'd run a single command (lj perhaps) and it would 1) compile the Less-Java file, exiting with error message(s) if there were compiler errors, 2) run any unit tests from the file, exiting with error message(s) if any of them fail, and then finally 3) run the program. This is similar to what the test script added in #34 does, and would make the TravisCI implementation of internal tests fairly easy. Then we can just write a simple bash function or script to "flip" the result for testing programs that are supposed to fail.
Also, this should probably involve hiding or obfuscating/renaming the generated files, both for simplicity and to enable multiple programs to be compiled, tested, and run in parallel (enabling faster testing). This was pointed out by @ripleymj in #37 (comment).
As mentioned in the discussion for #30, the README will need to be revised after we address this issue and those changes should be a part of any pull request to close it.
Activity
lam2mo commentedon May 29, 2019
Just as a note, I believe the original vision from Zamua's thesis is that you'd run a single command (
ljperhaps) and it would 1) compile the Less-Java file, exiting with error message(s) if there were compiler errors, 2) run any unit tests from the file, exiting with error message(s) if any of them fail, and then finally 3) run the program. This is similar to what the test script added in #34 does, and would make the TravisCI implementation of internal tests fairly easy. Then we can just write a simple bash function or script to "flip" the result for testing programs that are supposed to fail.lam2mo commentedon May 29, 2019
Also, this should probably involve hiding or obfuscating/renaming the
generatedfiles, both for simplicity and to enable multiple programs to be compiled, tested, and run in parallel (enabling faster testing). This was pointed out by @ripleymj in #37 (comment).lam2mo commentedon May 30, 2019
As mentioned in the discussion for #30, the README will need to be revised after we address this issue and those changes should be a part of any pull request to close it.
lam2mo commentedon Jan 24, 2020
@MrBartWolf owns this now. :)