Allowing header attributes

[olyerickson]

I was very excited to see strcode!

My team would like to be able to associate semantic attributes with blocks of code. I think this could be done using Markdown Extended's header attributes; see esp

http://rmarkdown.rstudio.com/authoring_pandoc_markdown.html

The idea would be to allow the specification of classes and attributes in the header via an extended version of the Shiny gadget. I'd like to start a discussion on this...

[lorenzwalthert]

ok, thanks @olyerickson for opening the issue. So I might need some further clarification on this to understand your request. So you would like to insert something like the following with strcode, right?

# _________________________________________
# my_header {#my_identifier .my_class} ####

Whereas the identifier and (multiple) attributes can be inserted via the Shiny Gadget. What is the ultimate goal of this? strcode was designed to work on plain .R files (or within junks of .Rmd files) and it does not support any markdown syntax. So if you use attributes and classes, I guess you would ultimately want to post-process the file with pandoc. Is that right?

[olyerickson]

Yes, that is precisely the case. We're working on something we're calling "semantic markdown," which we see as a variation of .Rmd that allows for associating semantics with code blocks.

The basic classes of the code blocks would be provided by a subset of ProvONE, http://vcvcomputing.com/provone/provone.html (Program, Subprogram, Data, Visualizations, etc). The headers would allow for additional classes (entities can have multiple types) as well as properties, etc.

We envision producing linked data like this http://orion.tw.rpi.edu/~olyerickson/cwf.ttl and a resulting graph like this http://bit.ly/cwf_rdf_visualization based on a script...

[olyerickson]

In our vision, "semantic markdown" renders to RDF or XHTML+RDFa (that's our problem). What's super interesting about strcode is that you provide a really nice interface for users to add code blocks, and an interface paradigm for specifying classes and properties. Very cool.

[olyerickson]

By our approach, it's possible to visualize the script structure e.g like this http://bit.ly/2ldwVEJ

[lorenzwalthert]

thanks for sharing this. I just watched your recent talk on semantic markdown which was also helpful to understand what you envision. I thought about how to extend strcode in that direction too (especially on the aspect of visualising dependencies and data analysis processes), but did not really make any progress. Hence, I would indeed be very happy to collaborate with you on that matter.
To this end, I have modified the "front-end" (and only the "front-end") of the Shiny Gadget (see new branch semantics), so one can now tick the box "add semantics", and three additional input fields open.

• A hash sequence is created automatically as an identifier.
• Classes must be from a given list (for now I specified some placeholder classes).
• key / value pairs must match the regex ^[a-zA-Z1-9]+\\s*=\\s*[a-zA-Z1-9]+$ in order to be considered as valid.

Feedback is most welcome, also on details. Please note that this is a skeleton only, I have some things in mind that could be improved as well:

• allow to specify the length of the unique identifier in the Shiny Gadget (for changing collision probability depending on size of a project and trading shortness vs. uniqueness)
• moving all buttons to the very bottom
• use better formatting (e.g. markdown) in the help pop-up.
• further restrict fields for identifier (must be unique within project, which might require expensive (in terms of time) checks of all headers in project / or shall we create an index? / how should we update manually removed headers?), regex (probably also some sort of uniqueness of key required)

As a next step, I suggest to implement a skeleton of the "backend", so the selections actually produce markdown compatible header attributes. If you want to, you are welcome to submit PRs if you want to code some of it yourself.

[olyerickson]

Thanks very much for this! I'll install the semantics branch and take a look

[lorenzwalthert]

I just pushed a version to the semantics branch that has a backend as well. Now, you should be able to insert with the Gadget what you wanted to. Feedback is most welcome.

[olyerickson]

Looking forward to testing it!

…

On Wed, Mar 15, 2017 at 5:15 PM, Lorenz Walthert ***@***.***> wrote: I just pushed a version to the semantics branch that has a backend as well. Now, you should be able to insert with the Gadget what you wanted to. Feedback is most welcome. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#5 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAV4-TIzTohHyZjTnkhwlpCJWVDxEdhOks5rmFTpgaJpZM4MPoSX> .

-- John S. Erickson, Ph.D. Director of Operations, The Rensselaer IDEA Deputy Director, Web Science Research Center (RPI) <http://idea.rpi.edu/> <olyerickson@gmail.com> Twitter & Skype: olyerickson

[lorenzwalthert]

Cool. It's not yet tested extensively.

[lorenzwalthert]

@olyerickson have you had a chance to test the semantics implementation?

[olyerickson]

I briefly did, and hope to get to it this week. The past few weeks have been grant writing ;)

…

On Sun, Apr 9, 2017 at 7:40 AM, Lorenz Walthert ***@***.***> wrote: @olyerickson <https://github.com/olyerickson> have you had a chance to test the semantics implementation? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#5 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAV4-Q3lvDA9vXXTppqx1_tICeXruHV3ks5ruMOkgaJpZM4MPoSX> .

-- John S. Erickson, Ph.D. Director of Operations, The Rensselaer IDEA Deputy Director, Web Science Research Center (RPI) <http://idea.rpi.edu/> <olyerickson@gmail.com> Twitter & Skype: olyerickson

[lorenzwalthert]

I was quite delighted to see @XiaoliangJiang working on an extension of strcode that was documented in his blog. I just had a look at his fork and I was wondering why all but 3 commits were gone. Is there any interest from your side to merge the changes back in the upstream branch? I think it would be nice to combine the functionality in one place.

[olyerickson]

Yes, we absolutely need to do this! You can see all of his code here https://github.com/XiaoliangJiang/semarkdown @XiaoliangJiang <https://github.com/xiaoliangjiang> Can you please respond? Thanks!

…

On Sun, Sep 24, 2017 at 5:24 AM, Lorenz Walthert ***@***.***> wrote: I was quite delighted to see @XiaoliangJiang <https://github.com/xiaoliangjiang> working on an extension of strcode that was documented in his blog <https://notebooks.dataone.org/semantic-annotation/>. I just had a look at his fork and I was wondering why all but 3 commits were gone. Is there any interest from your side to merge the changes back in the upstream branch? I think it would be nice to combine the functionality in one place. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#5 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAV4-QQe3_WmlOE6NnrgdHN889O-HRlzks5slh_rgaJpZM4MPoSX> .

-- John S. Erickson, Ph.D. Director of Operations, The Rensselaer IDEA Deputy Director, Web Science Research Center (RPI) <http://idea.rpi.edu/> <olyerickson@gmail.com> Twitter & Skype: olyerickson

[lorenzwalthert]

Cool. Looking forward to it. Will keep an eye on semarkdown. Let me know if there is anything I can help you out with.