Before this commit, const_get with inherit=true and constant lookup
expressions searched the ancestors of the starting point in an order
different from `starting_point.ancestors`.
Items in the ancestry list introduced through prepend were searched
after searching the module they were prepended into. This oddity allowed
for situations where constant lookups gave different results even though
`starting_point.ancestors` is the same.
Do the lookup in the same order as `starting_point.ancestors` by
skipping classes and modules that have an origin iclass. The origin
iclass is in the super chain after the prepended modules.
Note that just like before this commit, the starting point of the
constant lookup is always the first item that we search, regardless of
the presence of any prepended modules.
[Bug #17887]
When running btest there is a crash when compiled with
RGENGC_CHECK_MODE=4. The crash happens because `during_gc` is not
turned off before `gc_marks_check` is called, causing the marking to
happen on the main mark stack instead of mark stack created in
`objspace_allrefs`.
... instead of exact matching. I'm now creating a built-in gem that
modifies Exception's error message, so the expectation value is changed.
IMO, it is good to check that did_you_mean suggestion is NOT added in
the uncorrectable case.
https://github.com/ruby/did_you_mean/commit/ebe88ec4d2
Previously, did_you_mean used `msg.end_with?(suggestion)` to check if
its suggestion is already added.
I'm now creating a gem that also modifies Exception's message. This
breaks did_you_mean's duplication check.
This change makes the check use String#include? instead of end_with?.
https://github.com/ruby/did_you_mean/commit/b35e030549
Previously, DidYouMean::Correctable#original_message did
`method(:to_s).super_method.call` to call the original to_s method by
skipping Correctable#to_s.
I'm now creating a gem that prepends another to_s method to NameError,
which confuses the hack. An immediate solution is to replace it with
`method(:to_s).super_method.super_method.call` to skip the two methods.
But it is too ad-hoc.
This changeset uses more extensible approach and allow a prepended
module to declare that they should be skipped by defining a constant
named `SKIP_TO_S_FOR_SUPER_LOOKUP`.
https://github.com/ruby/did_you_mean/commit/8352c154e3
https://github.com/ruby/irb/pull/249 actually slowed down how `code` is
concatenated. The original way of creating `code` is faster.
[before]
user system total real
2.420137 0.005364 2.425501 ( 2.426264)
[after]
user system total real
1.000221 0.007454 1.007675 ( 1.008295)
Theoretically, this implementation might skip lines that don't appear in
Ripper tokens, but this assumes such lines don't impact whether the code
passes compilation or not. At least normal blank lines seem to have an
`on_ignored_nl` token anyway though.
https://github.com/ruby/irb/commit/27dd2867cd
When using a non-ASCII compatible source and destination encoding
and xml escaping (the :xml option to String#encode), the resulting
string was broken, as it used the correct non-ASCII compatible
encoding, but contained data that was ASCII-compatible instead of
compatible with the string's encoding.
Work around this issue by detecting the case where both the
source and destination encoding are non-ASCII compatible, and
transcoding the source string from the non-ASCII compatible
encoding to UTF-8. The xml escaping code will correctly handle
the UTF-8 source string and the return the correctly encoded
and escaped value.
Fixes [Bug #12052]
Co-authored-by: Nobuyoshi Nakada <nobu@ruby-lang.org>
Samuel Williams
Nobuyoshi Nakada
Keiko Kaneko
git
Peter Zhu
Takashi Kokubun
Masataka Pocke Kuwabara
Yusuke Endoh
Alan Wu
Yuki Nishijima
xtkoba
Jeremy Evans