Hi Elijah,
On Mon, Nov 4, 2019 at 8:14 AM Fossies Administrator <Jens.Schleusener@xxxxxxxxxxx> wrote:
Hi Elijah,
On Mon, Nov 4, 2019 at 7:07 AM Fossies Administrator
<Jens.Schleusener@xxxxxxxxxxx> wrote:
Hi,
although misspelling corrections are not the most exciting issues and the
spelling errors are rarely true code bugs but mostly contained in the
comments and documentation parts they correction may still improve the
overall quality of a software project a little bit.
In this sense I created a code misspelling report for "git" using the
program "codespell"
https://fossies.org/linux/misc/git-2.24.0.tar.xz/codespell.html
or version independent
https://fossies.org/linux/misc/git/codespell.html
Cool, thanks for sending this report along. The typos within the
Documentation/ subdirectory have mostly been addressed by the
en/doc-typofix branch (in next, not yet merged to master). There are
also some false positives in this report (e.g. mmaped should not be
changed to mapped, CREAT should not be changed to CREATE, examples in
format-patch showing how to correct spelling errors need to keep their
spelling errors or it won't make sense, and perhaps some others), but
most of them look like actual spelling errors that should be
corrected. I'll send in a patch, and mark you as the reporter of the
issues.
So, I used your codespell program
That seems to be a misunderstanding: I'm not the author of the codespell
program but I only use that program to detect spelling errors and point to
their existence while offering the option to inspect the context of the
probably misspelled words in a fast and comfortable way via a Web page.
to catch all these and turned
en/doc-typofix into a series of patches to fix all the errors:
https://public-inbox.org/git/pull.418.v2.git.1572973650.gitgitgadget@xxxxxxxxx/
But I thought it might also be worthwhile to you to report what the
false positives found by that program were; I've included them at the
end of this email in the form of a patch. The places where the program
seemed to struggle were:
* In dealing with translation files. It didn't recognize them as
such and often tried to translate foreign words to a nearby English
one.
* In handling variable names: acronyms might be similar to english
words (cas, for compare and swap, looks like case), abbreviations
might look like alternate words (ans, short for answer, looks like
and).
* Testcases with intentional spelling errors
* Proper names that were similar to English words (Ned -> Need,
Claus -> Clause)
* miscellaneous tech jargon or package names (e.g. 'filetest' module
being replaced with 'file test', 'ith' as in not first or second
but the item at position i being replaced with 'with', 'mmaped'
being replaced with 'mapped', 'CREAT' changing to 'CREATE',
'UserA' (out of a sequence of UserB, UserC, etc.) changing to
'users', 'spawnve' function name being replaced with "spawn",
'CAs' (certificate authorities) being replaced with 'case', etc.)
Thanks for the detailed and informed feedback. These are exactly problems
that I noticed as well. Some additional ones are for e.g. words at the end
of a string definition (apostrophe), mail addresses and differences
between US-English and UK-English. Sometimes it's difficult to decide what
to exclude: For e.g. your mentioned word "ans" is often intentionally used
but often it's also a typo ("and") so FPs or FNs seem unavoidable. So
codespell respectively Fossies can give only the pointers and an
individual check seems always required.
Some of the according FPs are excluded by Fossies generally, some other
obvious FPs are excluded by Fossies specifically for each FOSS project
(see always the bold item "Codespell configuration" with a link to
"Project-specific additions" or to "(no project-specific adaptions yet
done)" that shows all the excluded words and directories/files).
Some of these would be difficult for any tool to deal with, but e.g.
recognizing translation files as such and ignoring them might be
interesting.
As one can see on the page
https://fossies.org/linux/misc/git-2.24.0.tar.xz/codespell_conf_info.html
there are already done some according attempts.
To let run codespell with an English dictionary about directories like
"translations" or "langmap" is probably in the most cases meaningless
(although in such directories may be contained also "English" source
code). So if one uses codespell manually one should use options like
"--ignore-words-list" and "--skip" (directories, files) and can look
optionally for a starting point to the values Fossies has used.
Anyway, thanks for the report and the pointer to the tool.
Although (hopefully) most of the FPs within your list are already excluded
by Fossies I will check the list to make Fossies perhaps even smarter.
-- 8< --
Subject: [PATCH] DO NOT MERGE: False positives from `codespell -w`
[... the very big list of FPs removed in this reply mail ...]
Regards
Jens
--
FOSSIES - The Fresh Open Source Software archive
mainly for Internet, Engineering and Science
https://fossies.org/
