Lints incorrectly treating `-`-prefixed throwing tests as ordinary objects

[yegor256]

The EO parser marks a truthy test attribute (++> name) by prefixing its
@name with +, and a throwing test attribute (--> name) by prefixing it
with - (see Suffix.attribute in eo-parser). Three lints know about the
first prefix only, so every throwing test is linted as if it were an ordinary
object: names/non-kebab-name.xsl:14, misc/sparse-decoration.xsl:17 and
tests/wrong-test-order.xsl:14 all filter on starts-with(@name, '+').

On eo-runtime of objectionary/eo this produces 236 bogus warnings out of
338: 98 non-kebab-name ("Formation name
-on-bytes-of-wrong-size-as-i16 must be in kebab-case" — the leading hyphen
is the marker, not part of the name), 95 sparse-decoration (a throwing test
such as 01-01-01.as-i16 --> on-bytes-of-wrong-size-as-i16 is a single-φ
formation by construction, exactly the case starts-with(@name, '+') already
excuses for truthy tests), and 43 wrong-test-order (the last ++> test in a
file is reported as misplaced because the --> tests that follow it are not
counted as tests).

The smallest fix is to widen the three predicates from
starts-with(@name, '+') to also accept -, ideally through one shared
eo:test-name() function in org/eolang/funcs/, since
misc/incorrect-test-object-name.xsl:14,
names/unsorted-named-attributes.xsl:14,
tests/unit-test-without-phi.xsl:14 and tests/unit-test-missing.xsl:14
carry the same assumption and will need it too.

[yegor256]

@volodya-lombrozo could you take a look when you have a moment? Happy to clarify anything in the report that is unclear.

[yegor256]

This is still reproducible with org.eolang:lints:0.4.0, so I believe the issue was
closed prematurely — the fix lives in #1136, which is still open. The three predicates
in the released jar are unchanged:

$ unzip -p ~/.m2/repository/org/eolang/lints/0.4.0/lints-0.4.0.jar \
    org/eolang/lints/misc/sparse-decoration.xsl | grep -o "starts-with(@name, '+')"
starts-with(@name, '+')

Running mvn compile over eo-runtime of objectionary/eo (after renaming every
truthy test to the can-/cannot-/accepts-/rejects-/stops-on- convention that
bad-test-name now requires, which brought bad-test-name down from 1320 to 0) leaves
exactly 253 warnings, and every single one of them is a throwing test being linted as
ordinary code:

lint	count	cause
non-kebab-name	106	-stops-on-bytes-of-wrong-size "must be in kebab-case"
sparse-decoration	102	a --> test is a single-φ formation by construction
wrong-test-order	44	the last ++> test is reported as misplaced because the --> tests after it are not counted as tests
unit-test-missing	1	directory.open has only --> tests

I verified each of the 253 individually: the 209 non-kebab-name /
sparse-decoration warnings all point at a source line holding a --> declaration,
and the 44 wrong-test-order warnings all point at the last ++> test of a file whose
next sibling is a --> test.

One more prefix worth folding into #1136: tests/bad-test-name.xsl:14 also filters on
starts-with(@name, '+'), so a throwing test may be named anything at all without
being flagged, while its truthy twin must carry one of the five prefixes.

[0crat]

@yegor256 Thanks for closing this issue! You've earned +12 points for this: +8 as a basis; +4 for resolving it in less than 12 hours. Your running score is +2721; don't forget to check your Zerocracy account too).

[yegor256]

@rultor release, tag is 0.4.1

[rultor]

@rultor release, tag is 0.4.1

@yegor256 OK, I will release it now. Please check the progress here.

[rultor]

@rultor release, tag is 0.4.1

@yegor256 Done! FYI, the full log is here (took me 6min).