Open nicobrandt opened 3 weeks ago
In eLab, text
will be HTML for most cases (default setting). But it can also be Markdown. I suggest using encodingFormat
on the Dataset (text/html
, text/markdown
).
Even if these would be included in the Crate, the corresponding URL would need to be updated somehow in the text.
In my case, any attached file has it's old name as alternateName
, it corresponds to the name of the file on the storage backend, which allows one to look for this value (some hash) in the main text and replace it with the new one, of the freshly uploaded file. Examples .eln should ideally include such a use case, with an image embedded in main text.
In eLab,
text
will be HTML for most cases (default setting). But it can also be Markdown. I suggest usingencodingFormat
on the Dataset (text/html
,text/markdown
).
Including a MIME type somewhere would be ideal. For datasets and comments, this should work, as long as ELNs know to treat this property accordingly. Files may have both a MIME type (describing the actual file content) and a description though, so I can see it being an issue there at least.
Including a MIME type somewhere would be ideal.
But encodingFormat is a MIME type!
Files may have both a MIME type and a description
How is the description clashing with encodingFormat??
What I meant is, if a file has a description specified as text
and a MIME type (describing the file contents) specified as encodingFormat
, then how would we specify the MIME type of just the description?
Oh I see. Like this then:
"description": {
"@type": "CreativeWork",
"text": "<p>This is a detailed description in HTML format.</p>",
"encodingFormat": "text/html"
},
This syntax would be nice, but unfortunately only text seems to be allowed as value for the text
property. Using description
like in your example would actually allow us to use https://schema.org/TextObject though, which could look very similar to your example, just with a different @type
. So we could potentially rethink the use of text
and description
within the different entities.
Can you remind me, why we use "text" at all if there is description? I forgot.
I am thinking to always convert upon eln-import the content into the encoding that pasta uses (md). As Nico said, it is error-prone but it makes the code easier. I might change this point-of-view in the future.
We once discussed (somewhere) to use description
for describing the node itself and text
for the actual description "content". Currently, in Kadi4Mat we only use the former to add some meta descriptions about additional export files that the user may choose to include in the Crate. But these are not really necessary, so switching from text
to description
everywhere would not be a problem for us.
Have we discussed yet on how to deal with different description formats? Specifically, I'm talking about the
text
property we use within datasets, files, comments, etc. In the current examples we have, one can find at least three different formats: HTML, Markdown and plain text. Generally, ELNs could either leave them as-is and hope for the best, or attempt to convert the text in a suitable format. However, such conversions can be error-prone, especially when the ELN would first need to detect the source format based on the contents.A related question, at least for Markdown, HTML, etc., is also how to deal with linked, ELN-internal images. Even if these would be included in the Crate, the corresponding URL would need to be updated somehow in the text.