Hi Gregor, On Wed, 2023-08-09 at 03:21 +0200, gregor herrmann wrote: > I like duck and run it after each package build. > Unfortunately typically I get output from the wild-guess check for > some strings on websites where duck tells me that the upstream > homepage or the Debian BTS or a well-known license is a "parked > domain of for sale", and this test is almost always a false positive. > > Current example, package rex: > > I: debian/copyright:62: URL: > https://www.apache.org/licenses/LICENSE-2.0: INFORMATION > (Certainty:wild-guess) > Curl:0 HTTP:200 No error > Website seems to be outdated, is probably a parked domain or for > sale. Please update your links! > Matching regular expression(s): > m/\breplaced with\b/i > > I: debian/control: Homepage: https://www.rexify.org/: INFORMATION > (Certainty:wild-guess) > Curl:0 HTTP:200 No error > Website seems to be outdated, is probably a parked domain or for > sale. Please update your links! > Matching regular expression(s): > m/\breplace .* with\b/i > > I: debian/upstream/metadata:URL: https://github.com/RexOps/Rex.git: > INFORMATION (Certainty:wild-guess) > Curl:0 HTTP:200 No error > Website seems to be outdated, is probably a parked domain or for > sale. Please update your links! > Matching regular expression(s): > m/\breplace .* with\b/i > > I: debian/upstream/metadata:URL: https://github.com/RexOps/Rex: > INFORMATION (Certainty:wild-guess) > Curl:0 HTTP:200 No error > Website seems to be outdated, is probably a parked domain or for > sale. Please update your links! > Matching regular expression(s): > m/\breplace .* with\b/i > > > Checking for "deprecated" (on upstream websites which document > functions) or "replaced (by|with)" doesn't make any sense IMO … > Please just remove tese tests …
There are a couple of different points I'd like to address in order to fix this issue. First, I agree with you, "replaced (by|with)" and "deprecated" are too generic not to trigger false positives. I'll be removing them from the list. Secondly, even if, as stated by the check certainty, the suggestion is at most a wild-guess, I would like to keep the test as it can still be useful to catch deprecated projets or links that moved on to another page. However, I want to have a way for users to filter the checks based on certainty. I'll be adding an option for that both in the cli arguments and the configuration file. Although, I'll keep the default to show all checks. Finally, the checks for obsoletes sites is currently at a certainty of wild-guess. I'll be bumping that to possible as, to the contrary of the parked test, its a list of well known deprecated sites, and virtually has no chance of false positive. Best, -- Baptiste Beauplat
signature.asc
Description: This is a digitally signed message part
