Skip to content

Latest commit

 

History

History
76 lines (61 loc) · 2.8 KB

details.adoc

File metadata and controls

76 lines (61 loc) · 2.8 KB
Raw

Details

Table of Contents

Basics

  • The usages directory should contain files with comment block tests in them. Subdirectories of usages are recursively searched for such files as well. These files will be used to generate files that can be executed as tests.

  • The directory housing the generated test files is named .janet-usages and lives within one’s project directory.

  • Test results and output (from stdout and stderr) are stored in a subdirectory of .janet-usages. The name of this subdirectory is generated each time and it should be very unlikely to lead to collisions.

Concrete Example

.
├── .janet-usages
│   ├── .1614856826-00d80fdb386c5068-judge-gen
│   │   ├── 0-grammar.jimage
│   │   ├── stderr-0-grammar.txt
│   │   └── stdout-0-grammar.txt
│   ├── grammar.janet
│   └── grammar.judge
├── my-src-dir
│   └── fun.janet
├── project.janet
├── test
│   └── usages.janet
└── usages
    └── grammar.janet

Once the test files have completed running, there should be a subdirectory in .janet-usages that contains test results and output. The name of this subdirectory is highly likely to be different each time. An example is .1614856826-00d80fdb386c5068-judge-gen.

  • Test results have a name like <n>-<filename>.jimage where <n> is a number and <filename> is obtained from a source file’s name. An example is 0-grammar.jimage.

  • A test output file containing stdout output has a name like stdout-<n>-<filename>.txt. An example is stdout-0-grammar.txt.

  • A test output file containing stderr output has a name like stderr-<n>-<filename>.txt. An example is stderr-0-grammar.txt.

Discussion

  • The generated directory (e.g. .janet-usages) will contain a copy of the source files from which the generated tests (e.g. grammar.judge) are created. The generated test files are modified forms of the original source files (e.g. grammar.janet) and may contain import forms referring to other source files. In order for these import forms to function properly, the original source is copied and thus (most ordinary) relative imports should continue to function in the context of the generated directory.

  • To clarify, the content of the source directory is recursively copied to the generated tests directory. As mentioned before, this is for the purpose of import forms functioning correctly.

  • The generated test files use the suffix .judge. This should be sufficient to prevent collisions between the names of source files (which should have extension .janet) and the names of generated test files.

  • Co-existence with other testing methods is something the author has an interest in. Please report any encountered issues along these lines.