Add basic unit tests #240
Add basic unit tests #240
Conversation
FedericoCeratto
force-pushed
the
unit-tests
branch
from
8114fbe to
0f25bd9
Compare
|
Isn't this accomplishing the same job as the end to end tests we had before? |
|
@Ph0enixKM these are unit tests (instead of integration tests). The purpose is to (quickly) test the exact generated output. |
FedericoCeratto
force-pushed
the
unit-tests
branch
from
0f25bd9 to
78a33e2
Compare
|
@FedericoCeratto I know what these do but what's the practical difference? Oh... instead of testing result we test the resulting bash code. But then we'd have to modify a lot of tests since we'll be now working on making the bash look more readable |
|
@Ph0enixKM Yes, the tests are meant to capture the generated shellscript and reflect improvements in readability. More importantly they would detect any unexpected change in the script layout. The amount of tests should remain very modest. |
| fn basic() { | ||
| comp( | ||
| r#" | ||
| import * from "std" |
There was a problem hiding this comment.
This import is unnecessary
| import * from "std" |
|
|
||
| /// Compile code and compare it with the expected value. | ||
| /// If any difference is found panic and show a friendly diff | ||
| pub fn comp(source: &str, expected: &str) { |
There was a problem hiding this comment.
Since comp can be read as compute or compare, can we change this name to a more verbose one?
This is just and example:
| pub fn comp(source: &str, expected: &str) { | |
| pub fn compare_bash_output(source: &str, expected: &str) { |
There was a problem hiding this comment.
maybe more like
| pub fn comp(source: &str, expected: &str) { | |
| pub fn compare_strings(source: &str, expected: &str) { |
There was a problem hiding this comment.
maybe more like
Sounds good too
| // Dedent code blocks based on the first line after `r#"` | ||
| let mut lines = expected.lines(); | ||
| lines.next(); | ||
| let line = lines.next().unwrap_or(""); | ||
| let indent = line.chars().take_while(|&c| c.is_whitespace()).count(); |
There was a problem hiding this comment.
Can we move this to a separate function?
| let compiled = compiled[FILE_HEADER.len()..] | ||
| .trim_end_matches("\n") | ||
| .trim_start_matches("\n"); | ||
|
|
||
| if compiled == exp { | ||
| return; | ||
| }; | ||
|
|
||
| let changeset = Changeset::new(&compiled, &exp, " "); | ||
|
|
||
| let mut out = String::new(); | ||
| let mut cnt = 0; | ||
| for diff in changeset.diffs { | ||
| match diff { | ||
| Difference::Same(ref x) => out.push_str(x), | ||
| Difference::Add(ref x) => out.push_str(&format!("\x1b[32m『{}』\x1b[0m", x)), | ||
| Difference::Rem(ref x) => out.push_str(&format!("\x1b[31m『{}』\x1b[0m", x)), | ||
| } | ||
| cnt += 1; | ||
| } | ||
| println!("--- {} differences found ---", cnt); | ||
| println!("{}", out); | ||
| println!("--- generated code ---"); | ||
| println!("{}", compiled); | ||
| println!("======================================="); | ||
| panic!("Unexpected difference"); |
There was a problem hiding this comment.
Can we also move out this part to a separate function?
|
@FedericoCeratto can you check this PR? |
| echo "Hi, I'm {name}. I'm {age} years old." | ||
| "#, | ||
| r#" | ||
| __0_count=1 |
There was a problem hiding this comment.
okay but this will be a pain in the ass to maintain.
compiled output should be expected to change as features rolled out, as well as there have been talks to simplify emitted variable names, remove awk and bc dependency and #72
There was a problem hiding this comment.
@b1ek the point is to have just a few of these tests. This way we know if something does not blow out of the water. But I agree that this is something that is more suited for the stable release
There was a problem hiding this comment.
so maybe we can do a minimum of tests?
Like for importing and echo?
|
@FedericoCeratto just a ping |
|
@FedericoCeratto @amber-lang/trusted-maintainers I'll close this PR for now so that it does not take space in open PR list. We can always reopen it |
This change adds basic unit tests and a test harness that compiles code and compare its with an expected value.
If any difference is found the test fail and show a friendly diff and the generated code.