Each msDesc display should contain link to the source XML here on github about medieval-mss HOT 12 OPEN

I was considering a process which would upload the catalogue GIF images to the server but keep them in this git repo (see #21). I could also add a section which would upload each of the TEI files to the server so that we can serve them out there, rather than point them to GitHub.

@holfordm which would you prefer?

from medieval-mss.

@ahankinson @IgnatG shall we go ahead with linking to the files on GitHub? If so how should we implement it in the HTML? How about, for example, a <div> at the bottom of the record which would contain (1) link to the TEI (2) details of licence for the record (cf. #56) (3) suggested citation (cf. #39).

• The TEI link would look something like the one here http://syriaca.org/person/1102 using the TEI logo.
• The licencing statement would say something like "metadata licence {name of licence with link}"
• I shall agree the citation with Martin but it would include slightly more information than the bare URL, including date accessed which presumably would be added by javascript

from medieval-mss.

Is it an issue that the version of the source XML here could be ahead of what was used to generate the HTML view on the production web site? Unless release branches are used, and there'd be no way for the XSL or the site to know which to link to. Or would users just want the latest?

from medieval-mss.

Possible way around this - don't know whether it would work in practice...
Future updates on production are versioned. Versions are recorded in the TEI <editionStmt> with a reference to the commit used. The XSL uses this to link to the correct version of the file on github.
Users may also appreciate (as well as the link to the raw TEI) a link for something link "view the history of this file".

from medieval-mss.

There's a chicken-and-egg problem here. The 'commit used' (the SHA1 hash key, which is what git uses to identify commits) in the editionStmt wouldn't be known until after the commit is made, so you would have to commit first to know it, then change it, then commit again, at which point the editionStmt is out of date...

Parsing out the git hash may also be difficult in XSLT.

I believe what Andrew M. is referring to is making the distinction between 'develop' and 'master' branches, and then using the git 'tag' functionality to identify a particular release. This way we understand that the files as they are on the master branch are the same that are live; any changes to the files between releases are done on the develop branch.

Personally, I'm all for this. However, I suspect it will be difficult to communicate and adhere to the same release strategy across all cataloguers.

from medieval-mss.

Hm, good point.
The links from the page could say something like:

1. TEI on GitHub (most recent version, may differ from the record displayed here)
2. TEI on GitHub (version history)

from medieval-mss.

Can we pick this up again in conjunction with the other catalogues? (I know that Fihrist at least were keen to have these links).

• is the master / develop branch idea feasible? (I suspect not unfortunately)
• probably the lag between live version and git version won't be significant as the catalogues will be reindexed fairly regularly (?) - but the potential lag should still be signalled
• what other information might go in the footer (?) with this link (e.g. license where this has been decided)

from medieval-mss.

In retrospect, I think my point about the TEI XML on GitHub being possibly ahead of what's indexed and displayed on the catalogue web sites, by a few days or possibly weeks, isn't worth worrying about. Probably not for individual manuscripts. Releases might be worth considering for the benefit of people downloading the whole collection, to run some analysis or text-mining, so they'd have a copy they can repeatably reproduce and cite when writing up their findings. You don't need branches to create releases, just pick a point in time when the files are stable. If that is at all likely, something could be added to the About page linking to this page on GitHub where releases would be listed if there were any.

from medieval-mss.

What is holding up adding a link to view the source is the need to schedule and prioritize the design work that would presumably be needed to add a TEI logo, and incorporate the other stuff requested in some kind of footer. Just adding an <a href="https://github.com/blah">View source</a> link could be done immediately (it would appear on a new line after the last piece of description currently, or just above the "Zotero bibliography" subheading for manuscripts that have been matched by that service.)

from medieval-mss.

-For the moment, then, shall we go with a plain link <a href="https://github.com/blah">View source XML on GitHub</a>? The TEI logo doesn't strike me as essential. We do need a licence statement, but AFAIK the library hasn't approved this yet, so it will need to be added later.
-Citation info: do we need the date of access to be added (which would require gabriel's time?) or could we do something like 'Bodleian Library, MS. Douce 62 (manuscript description), Medieval Manuscripts in Oxford Libraries, https://medieval.bodleian.ox.ac.uk/catalog/manuscript_4731 {date accessed}"
-It's a slight nuisance that it will appear before the Zotero bibliography; is there any way round this?
-I think implementing releases for the benefit of anyone wanting to use the whole dataset is an excellent idea; I'll open up a separate issue relating to that.

from medieval-mss.

I'll implement a simple view-source link in Fihrist, then let you know so you take a look at it. Possibly the Zotero bibliography could be moved with CSS, but that would be a job for Gabriel.

from medieval-mss.

Is there a way of capturing the hash and date of the latest Git commit during the indexing process (either by asking for it in the terminal or fetching it from GitHub with the XQuery?)
Could that information then be used to ensure that links to the source XML did in fact point to the correct version? And additionally to generate a citation reference that included the hash of the relevant commit and/or its date?

from medieval-mss.