Re: [Cgi-session-user] JSON problems as per RT bug 25325
Brought to you by:
sherzodr
From: Mark S. <ma...@su...> - 2008-03-11 14:12:14
|
Ron Savage wrote: > Hi Mark > > (1) JSON > > I replaced JSON::Syck with JSON::XS to see what would happen, but the > test program t/g4_dbfile_json.t still dies horriby, but differently. > > JSON::Any just tries to load JSON modules one at a time, so it would > stumble across either of the above 2 if they were installed. So it > offers no advantage. > > A patch will remove JSON from the docs, tests, and serializers. Great. Thanks! > (2) YAML > > I do not get the error reported: > http://rt.cpan.org/Public/Bug/Display.html?id=18948. > > My versions are: > ron@zoe:~/perl.modules/Patched-4.20$ perl -MYAML -e 'print > "$YAML::VERSION\n";' > 0.66 > ron@zoe:~/perl.modules/Patched-4.20$ perl -MYAML::Syck -e 'print > "$YAML::Syck::VERSION\n";' > 1.04 > > So what's the policy? Do we let uses have installation failures until > they upgrade, or do we chop YAML from the list of supported serializers? > > I see a problem which chopping YAML (and JSON): People who upgrade to > 4.21 (to be released with a dozen patches) and who are using YAML will > lose support for YAML in order to gain the advantages of the patches. I have used the YAML driver at least once myself and some interest in seeing it maintained. I mind less that it is in the core. The Makefile.PL could declare YAML as a feature. If you choose it, you get the latest versions of YAML. Still, I could see how breaking it out as different distribution could be cleaner, and could still be an optional feature. I don't think removing JSON or YAML in a future version would cause breakage for people who upgrade, because as far as I'm aware Perl installations don't delete files from previous versions, so the JSON and YAML drivers would remain on disk. It would only become a problem on an upgrade via "clean install". In sum, as long as you are working on it Ron it's fine with me to steer the solution towards your own inclinations, but I'm happy to be a sounding board if you are uncertain about what seems best. Mark |