This is blocking deployments right now.

Between vagrant and debian repos and some work with docker, I am going to mark this resolved. The broader packaging and 3rd party support discussions are happening elsewhere.

Let us review output to see if there is anything else left in output that can be cleaned out here and resolve this ticket.

This was removed as part of T55432: Typeof cleanup and smart serialization of new language and interwiki links. With ISBN, RFC, etc links, it is possible for clients to figure this out, but this is far more complex for interwiki links. So, we should consider adding this back since this affects HTML clients that inspect Parsoid HTML (for semantics or for styling in UIs).

In T103636#3614315, @Deskana wrote:

@ssastry Is this something your team could take a quick look at to gauge the difficulty of a fix? Thanks. :-)

Of the 491 pages from mediawiki that went through visual diff testing, Here is the list of 160 pages that had non-perfect visual diffs. Only 16 of those have a score about 1000 which means, they were not pure vertical shift whitespace differences. I looked at many of those pages, and it is hard to actually tell the rendering difference. So, that is actually a good sign that once high priority errors are fixed, visual diff numbers are more conservative than what the actual result will be when Tidy is replaced.

In T165520#3607289, @akosiaris wrote:

I think we are done. The new parsoid boxes are up and running. They are running Debian stretch and nodejs 6.11. They are not pooled and do not serve any kind of traffic currently. Icinga is green and seems like all checks are running fine. @ssastry @Arlolra, would you like to test the new boxes and give a thumbs up (or down...) ?

Clearly an infinite loop of some sorts

[subbu@earth mediawiki] parse.js --prefix afwiki --page 'Sjabloon:Omreken/Duaal/KafAafVhEaf' < /dev/null
[fatal][afwiki/Sjabloon:Omreken/Duaal/KafAafVhEaf] Maximum call stack size exceeded
RangeError: Maximum call stack size exceeded
    at ParsoidLogger.Logger.log (/home/subbu/work/wmf/parsoid/lib/logger/Logger.js:103:3)
    at AsyncTokenTransformManager.emitChunk (/home/subbu/work/wmf/parsoid/lib/wt2html/TokenTransformManager.js:377:12)
    at TokenAccumulator._callParentCB (/home/subbu/work/wmf/parsoid/lib/wt2html/TokenTransformManager.js:1322:16)
    at TokenAccumulator.receiveToksFromChild (/home/subbu/work/wmf/parsoid/lib/wt2html/TokenTransformManager.js:1253:7)
    at AsyncTokenTransformManager.maybeSyncReturn (/home/subbu/work/wmf/parsoid/lib/wt2html/TokenTransformManager.js:835:3)
    at TokenAccumulator._callParentCB (/home/subbu/work/wmf/parsoid/lib/wt2html/TokenTransformManager.js:1322:16)
    at TokenAccumulator.receiveToksFromChild (/home/subbu/work/wmf/parsoid/lib/wt2html/TokenTransformManager.js:1253:7)
    at AsyncTokenTransformManager.maybeSyncReturn (/home/subbu/work/wmf/parsoid/lib/wt2html/TokenTransformManager.js:835:3)
    at TokenAccumulator._callParentCB (/home/subbu/work/wmf/parsoid/lib/wt2html/TokenTransformManager.js:1322:16)
    at TokenAccumulator.receiveToksFromChild (/home/subbu/work/wmf/parsoid/lib/wt2html/TokenTransformManager.js:1253:7)

Related WIP document: https://www.mediawiki.org/wiki/Parsing/Notes/Wikitext_2.0/Typed_Templates

In T175452#3593820, @Izno wrote:

Excellent, thanks. We'll talk it over onwiki about what to do.

You will see on https://en.wikipedia.org/wiki/Streptomyces_carpinensis that the listing under binomial name is not in bold at all. Only the italics is preserved. The two sets of bold tags are cancelling each other out.

Let us take a look at [[Streptomyces carpinensis]]. I effectively run it through expandtemplates API call and see this in the output '''<span class="binomial">'''''Streptomyces carpinensis'''''</span>''' which comes because of the template parameter | binomial = '''''Streptomyces carpinensis''''' This generates <b><span class="binomial"></b><i>Streptomyces carpinensis</i><b></span></b> which is really a case of bad nesting. Parsoid was unable to recognize it as misnested tags and is instead reporting it as pair of linter errors (stripping </span> tag and a missing </span> end tag).