Can't put html formatting into dc.description (or other fields) about documentation

[whikloj]

Yeah, I did the same thing. I think it is getting encoded either when it is sent to Fedora or on the way back.

@daniel-dgi suggested that we could probably fix this in the json -> json-ld conversion.

But you are correct, there is the question if we should allow this or not.

[philred]

I think this gets filtered out here:

https://github.com/Islandora/islandora/blob/7.x/includes/dublin_core.inc#L137

[whikloj]

@philred++

[rosiel]

@philred I don't think that's the problem because
a) if you don't specify otherwise, filter_xss() permits 'em' tags, and
b) that code is in Islandora/islandora, i.e. it's the 7.x-1.x branch.

[ruebot]

@philred, that link is to the current Islandora 7.x-1.x code base. So, it wouldn't apply here 😄

[bradspry]

After reading every DC standards document I could find, I would say the spirit of DC is plain text.

Based upon that, one can confidently say HTML in DC would be non-standard.

But clearly, standards never stopped anyone... For example, colon is a reserved character in URLs, but people still use colons every day in URLs by escaping them (or have the browser do it for them automagically)

There is no doubt in my mind, if one has the will, one can find a technical way to make it happen. For example, by wrapping HTML in CDATA.

But maybe it's a non-technical, Jurassic Park-type question: just because you can do it, should you? Again, that question doesn't stop most people, especially wildly talented people like Islandorians :)

Who am I to say no, or that it's not a good idea, when someone has a need to format their plain-textual data? A very common use case is simply wanting to add a new line to break up paragraphs. It's really obvious when you see it; I'm sure you see it every day.

I guess another perspective is machine-readable vs. human readable. Is there a happy, middle ground that can be found, that would be acceptable to all of the various components the mixed formatted text will pass through?

[dmoses]

Just to note it ... there are schema that do allow markup within elements. EAD is one - the bioghist element allows a number of tags. DDI 3.2 is another schema we use that allows for XHTML elements (see the DDI User Guide - Appendix B: XHTML Tags Support by DDI ).

[rosiel]

I'm fine with enforcing that DC be plain text (by whatever transform is currently going on) so long as there is a way to write, store, and access marked-up text in another field - like Donald pointed out, there are some places where that would be in-spec. The user need (as expressed to me, which resulted in some unpleasant hackery) is that sometimes text is grammatically wrong unless markup (italics, super/subscript, etc) is present. This is especially true in languages with prescriptive rules such as French.

[dannylamb]

Clearing out D7 issues.