`beaker-wizard` requires author name to consist of letters in ASCII printable range - basically, lowercase and uppercase latin letters (a-z). In 2017 there is hardly a technical reason for such limitation. Version-Release number of selected component (if applicable): $ beaker-wizard --version beaker-wizard 2.3.0 How reproducible: Always Steps to Reproduce: 1. Run: beaker-wizard -n 'Gęśla Jaźń' -d "Issue reproducer" -t 5m -b CoreOS/beaker-client/Interoperability/sample-test-name 1469706 Actual results: #v+ Contacting bugzilla... Fetching details for BZ#1469706 Author's name ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [Gęśla Jaźń is not a valid author, what about Mirosław Zalewski?] #v- Expected results: Test case details are displayed for review. Names with diacritics or these not based on latin alphabet are accepted. Users are not told their real names are "not valid".
Created attachment 1325914 [details] Very simple patch to solve this issue (untested)
I would go one step further and say there is no reason why we should be requiring that people's names be either two or three space-separated elements, either. Possibly that is designed to catch the cause where someone's name has wrongly been guessed as their username (like "dcallagh") although if someone really wants to use that as their name, I don't see why the wizard should stop them. However, we might have a problem one layer down, in the testinfo.desc parser. I tried putting a non-ASCII name into a task Makefile and I get: rhts-lint testinfo.desc testinfo.desc:1: error: "Gęśla Jaźń <dcallagh>" is not a valid Owner field (should be a valid RFC2822 name_addr, such as John Doe <jdoe>) make: *** [testinfo.desc] Error 1 so that might need fixing too...
So, this was surprisingly hard to fully fix. :-) https://gerrit.beaker-project.org/5885 beaker-wizard: don't enforce any restrictions on author name https://gerrit.beaker-project.org/5884 accept non-ASCII names in testinfo.desc https://gerrit.beaker-project.org/5886 accept non-ASCII names in testinfo.desc
The random smattering of .decode() in testinfo.py makes me nervous though. So I'm thinking I should audit that whole module for unicode-cleanliness first, and then these patches can go on top. Bumping this to 25.0 since it's not really a minor fix.
I feel more confident now with this: https://gerrit.beaker-project.org/5904 testinfo.py: deal exclusively in unicode not bytes https://gerrit.beaker-project.org/5884 accept non-ASCII names in testinfo.desc and matching changes on the Beaker side: https://gerrit.beaker-project.org/5905 testinfo.py: deal exclusively in unicode not bytes https://gerrit.beaker-project.org/5906 accept non-ASCII names in testinfo.desc Although now that I think about it, the Beaker-side patches should really include a test case with a task with non-ASCII chars in the name...
A test on testinfo.desc with an owner name with special characters produces this error > rhts-lint testinfo.desc testinfo2.desc:1: error: "Gęśla Jaźń <bpeck>" is not a valid Owner field (should be a valid RFC2822 name_addr, such as John Doe <jdoe>)
Turns out my patch: https://gerrit.beaker-project.org/c/5904/ was not complete. rhts-lint and rhts-mk-generate-specfile do their own file handling which is currently still dealing in bytes and not unicode.
(In reply to Dan Callaghan from comment #8) https://gerrit.beaker-project.org/5967
This bug fix is included in rhts-4.73-1.git.4.8daecb4 which is currently available for download here: https://beaker-project.org/yum/client-testing/
Verified that * with beaker-client-25.0-0.git.176.b72276ac3.el6eng.noarch.rpm installed * Running: `beaker-wizard -n 'Gęśla Jaźń'` No longer produces the error "not a valid author", and produces a Makefile with the provided author name. * Running: `make rpm` in the directory with the makefile builds successfully and * Running: `rhts-lint testinfo.desc` Does not throw any errors
Beaker 25.0 has been released. Release notes are available upstream: https://beaker-project.org/docs/whats-new/release-25.html