Download a smaller list of bugs from W3C's Bugzilla

[foolip]

Limit the search to bugs that have a URL and are in interesting
components.

This shrinks the downloaded w3cbugs.cvs from 3211 bugs (242 kB) to 265
bugs (40 kB), with corresponding savings in transfer time both when
downloaded and then uploading to the wattsi build server.

This removes the following HTML Checker and SVG bugs from the output:
https://www.w3.org/Bugs/Public/show_bug.cgi?id=13648
https://www.w3.org/Bugs/Public/show_bug.cgi?id=13649
https://www.w3.org/Bugs/Public/show_bug.cgi?id=16125
https://www.w3.org/Bugs/Public/show_bug.cgi?id=23157
https://www.w3.org/Bugs/Public/show_bug.cgi?id=23165
https://www.w3.org/Bugs/Public/show_bug.cgi?id=23753

Based on findings from whatwg/html#619

[domenic]

So is the idea that those bugs used to be in the margins, but don't really make sense to track in HTML?

[domenic]

What is &bug_file_loc=http&bug_file_loc_type=substring about?

[foolip]

That's the "bugs that have a URL" bit from the description. The URL field is bug_file_loc and I wanted it to be non-empty since that's what wattsi uses to match bugs to sections. Finding no direct was of checking for non-emptyness, I just checked if it contained "http" instead, which yields the intended results.

[foolip]

Right, I grouped the bugs by product in whatwg/html#619 and found that in the HTML Checker and SVG products, none of the bugs are about the HTML spec or otherwise actionable for us. For reference, here's the list as formatted in whatwg/html#619:

HTML Checker

• Validation of image(s) that form(s) a link or button
• Implement JavaScript syntax checking in <script>
• A more informative message for “obsolete permitted DOCTYPE”
• XHTML: validator doesn’t check that Raw Text elements (style/script) match the constraints

SVG

• SVG should have image role
• Expose SVG <title> and <desc> to AAPIs

[foolip]

(It would have been easy to include them with product=HTML%20Checker&product=SVG.)

[sideshowbarker]

About the HTML Checker bugs, anyway I’m not encouraging people to use that any more but instead steering new bug reports to https://github.com/validator/validator/issues and I guess should just close the component to new bugs and move the existing bugs out from bugzilla and into that github tracker. And regardless, going forward I will ensure that no bugs get added to that component in a way that will affect the HTML spec annotations.

[domenic]

Looks good, merging. Eventually we need to find way to start adding GitHub issues into the sidebar like we do legacy bugs.....

[foolip]

#79

[foolip]

@sideshowbarker, the HTML Checker bugs aren't downloaded any longer, so no need to do anything further for the HTML spec's sake.