Three R packages cannot read the files created by xlsx about xlsx

[qrilka]

@stla I think we have talked about using more standard namespace prefixes - I don't have any objections against it - it just means R library authors can't write XML parsing code correctly :)

[stla]

They can, since the current released version of readxl can read the file.
Yes, we already talked about the vt namespace, but actually it was an issue with the Javascript library js-xlsx.

[jennybc]

I assure you we R package authors understand XML parsing and namespacing.

In readxl, we wrap an external library, RapidXML, for it speed. Unfortunately, it does not genuinely support XML namespaces and we will have to patch it or workaround that some other way for xlsx that uses nonstandard namespace prefixes. Which seems to consist substantially of xlsx files written by this Haskell library. However this is not the only source and, of course, I want readxl to read everything it should. It's just a matter of prioritization.

So we are working under constraints and making compromises, not incompetent.

Unless there's a specific advantage to using nonstandard namespace prefixes, it would seem to make sense to use the same prefixes as Excel itself, since most xlsx readers will be optimized for that.

[qrilka]

@stla could you check out 0ebcc90 ?
It fixes only 1 prefix in only 1 file - I guess that could be enough at least for your case.
If it is then I could release this quickfix as 4.3 alongside with sheet protection which is in the master already and with upcoming fix for #77

[stla]

Hi @qrilka
I've just tried.

• readxl: I didn't manage to install the development version, so I can't say (permission denied to copy a dll - the laptop of my job company has a bunch of such funny features). I'll try later, maybe I need to delete the previous version.

• tidyxl: no change, reading the file still crashes the R session

• openxlsx: it still cannot read the file, but the error message is different:

> openxlsx::read.xlsx("db0a04c.xlsx")
Error in read.xlsx.default("db0a04c.xlsx") : expecting a single value

[stla]

I finally managed to install readxl (perhaps I didn't need to delete the previous one, actually I had another opened session of RStudio).
And yes, it works !!

I will send the file to @nacnudus for tidyxl.

[stla]

@qrilka
To conclude, your update is nice. The failure of openxlsx is not highly unfortunate, since we rather use this package for writing (readxl is better for reading).

[qrilka]

Closing this as resolved, if there is still something missing - pleas raise your voice