On Sat, Oct 14, 2023 at 09:41:46AM +0300, Eli Zaretskii wrote: > > Eli, are you able to test this from git or do you need me to make another > > pretest release? > > Git is a bit problematic, as some of the file names include non-ASCII > characters. For this reason, and also for others (e.g., I have > already made too many changes to 7.0.93 sources), I'd prefer another > pretest.
The test suite tests file names with non-ASCII characters for the purpose of testing encoding issues and better supporting use of different character encodings. However, this has the ironic effect of making the Texinfo source code *less* accessible, especially on MS-Windows, or any system where non-ASCII bytes in file names downloaded from git cause an issue. It seems that these systems are the ones where character encoding problems would most likely occur, so we would want users to be testing and reporting any such problems. Running $ LC_ALL=C find . -name '*[^ -~]* to find files with non-ASCII names, all the results are under tp/tests. Hence, it would be straightforward to eliminate such file names, with the cost of also eliminating these tests. Since these tests contain these characters in reference output file names, there would seem to be no simple way to adjust them to work with fully-ASCII file names. (The only possibility to me appears to be to store them in git in some kind of coded form, and then to decode them when the tests run. For example, as a tar archive file with an ASCII file name. This would be some work to implement, of course.) For example: $ find ./tp/tests/encoded/res_parser/non_ascii_test_epub ./tp/tests/encoded/res_parser/non_ascii_test_epub ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8.2 ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8.1 ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/xhtml ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/xhtml/osé_utf8.xhtml ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/xhtml/Chapteur.xhtml ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/xhtml/nav_toc.xhtml ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/images ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/images/2-an_image.png ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/images/1-an_image.png ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/osé_utf8.opf ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/mimetype ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/META-INF ./tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/META-INF/container.xml The question is how beneficial it would be to have wholly ASCII file names for all files tracked in git. By default, we will keep these tests, but could decide otherwise if there is feedback from users of systems where they cause an issue.