can't use a non-prefix?

[mdsumner]

I'm not sure my understanding is correct, but I want to avoid the "d1" prefix and not have one:

uri <- 'https://dapds00.nci.org.au/thredds/catalog.xml'

thredds::CatalogNode$new(uri, prefix = "")$prefix
#> [1] "d1"
thredds::CatalogNode$new(uri, prefix = NA_character_)$prefix
#> [1] "d1"
thredds::CatalogNode$new(uri, prefix = NULL)$prefix
#> [1] "d1"
thredds::CatalogNode$new(uri, prefix = character(0))$prefix
#> [1] "d1"

?

[eblondel]

Hi @mdsumner initially the prefix had to be set manually, the reason why we still have it as legacy arg.

At some point I wrote some code then to inherit it from the xml namespace definition, no need to declare this prefix from user viewpoint. The prefix for 'http://www.unidata.ucar.edu/namespaces/thredds/InvCatalog/v1.0' namespace is inherited from xml2; If no specific (local) xmlns is given by Thredds, xml2 sets it to 'd1', that's the default behavior of xml2.

IMHO we should get rid of the prefix argument in the different functions, we just need it to perform the relevant Xpath behind the scene. There is no need as well to handle the namespace prefix as public field, and through the print method
Let's see what @btupper thinks of this.

[btupper]

@mdsumner your example reminds me of Henry Ford's quote ...

Any customer can have a car painted any colour that he wants so long as it is black.

I can't recall the circumstances that drove the need for handling prefix, and I would defer to @eblondel to guide us to the right solution which is to either remove the optional prefix handling as suggested or to fix it so the user can override the default.

If it the choice is to fix the behavior then I think the only place we need to address it is here. We would just have to test if the prefix argument is NULL before assigning the value.

Maybe something along the lines of ...

if (is.null(prefix)){
  self$prefix <- names(ns)[ns == "http://www.unidata.ucar.edu/namespaces/thredds/InvCatalog/v1.0"]
} else {
  self$prefix = prefix[1]
}

But if it really isn't necessary at all for the user to optionally manage it then I am fine with removing its exposure.

[mdsumner]

ah ok, I need to explore more - but the immediate motivation is I couldn't get it to work with this thredds, so I picked on the first thing that seems to a blocker , and maybe that's not the real issue 🙏

[btupper]

In the meantime... might you try the fix-prefix branch and see it provides a short term solution?

remotes::install_github("BigelowLab/thredds", ref = "fix-prefix")

[mdsumner]

that works:

thredds::CatalogNode$new(uri, prefix = "")

thanks, I'll keep exploring

[mdsumner]

just fyi I can't make any sense of the xml I was trying to explore with this package, it doesn't work at all and I think their index is a bit broken, doesn't seem to be a problem in this package.

I will explore their stuff from the inside 🙏