Return a POSIXct timestamp instead of a non-standard locale dependent date

[jmaspons]

No description provided.

[mpadge]

@jmaspons Thanks for picking this one up too. There was a reason for that originally, which I unfortunately forget the exact details of. It was related to what happens when writing and re-reading an XML doc via osmdata_xml(). Can you maybe just add a test to demonstrate that your updates here are successfully carried across with a workflow like this:

f <- "doc.osm"
osmdata_xml(q, filename = f)
x <- osmdata_sf(q, doc = f)

❓

---

That makes me also realise that we've got an inconsistency there, with filename in osmdata_xml, and doc in the read functions (_sf/_sc). Looking at the git blame, it seemed that was introduced in 920fae2. We should make the nomenclature consistent there, which would be a potentially breaking change that would require a deprecation step.

[mpadge]

I'll request changes here, just so I see your response to #416 (comment). Thanks!

[jmaspons]

I added tests to compare timestamps for results from queries and from doc.

For the filename / doc, it has been always filename for osmdata_xml() (parameter introduced in fdf15f0), but we can change to doc and deprecate filename. I'm not convinced by the doc naming better xml or data, but it works and there is no need to deprecate all the osmdata_*(). I'will do it in another PR.

[mpadge]

@jmaspons I don't mind at all whether they're called doc, filename, or even something else, but just thought it would be good to be consistent. Then again, filename names an output file, whereas doc names an input file. That might be an argument for having different names and ignoring what I said anyway. It was just a thought; happy to leave up to you. thanks 👍

[mpadge]

Great, thanks!