Glossary returning unanticipated definitions, if leading characters match

[michaelmcd18]

Found this case while using the new read_glossary() function. Here's the example case:

The glossary.tex file contains definitions for ng and MCMC. It does NOT have definitions for n or MC. However, I get the following:

> glossary$n
nanograms (ng)
> glossary$MC
Markov chain Monte Carlo (MCMC)

It seems if the leading characters match an existing definition it returns that definition. I'd expect an error if there wasn't an exact match.

[kylebaron]

Thanks for the report, @michaelmcd18. Lists do partial matching by default. I can turn this off for glossary objects.

x <- list(michael = "scientist")

x$mi
#> [1] "scientist"

^{Created on 2024-06-12 with reprex v2.0.2}

[kyleam]

[ I see @kylebaron just posted, but I'll post this reply just for additional info. I'm not against turning it off. ]

Related to discussion at #326 (comment)

It seems if the leading characters match an existing definition it returns that definition. I'd expect an error if there wasn't an exact match.

Hmm, I'm not a fan of the partial match with $ (and I like tibble's departure from it), but I don't think it's altogether surprising given the standard $ behavior:

x <- list(foo = "foo", bar = "bar")

x$fo
#> [1] "foo"

options(warnPartialMatchDollar = TRUE)
x$fo
#> Warning in x$fo: partial match of 'fo' to 'foo'
#> [1] "foo"

One way to avoid it is with [[ and then a check for null:

x[["fo"]]
#> NULL

[kylebaron]

I took care of this in yspec, which is similarly a list of lists; can handle it the same way here.

https://github.com/metrumresearchgroup/yspec/blob/main/R/class-yspec.R#L78-L84

[michaelmcd18]

Thanks for the explanations @kylebaron @kyleam! I certainly don't want to add extra work, it seems like if I was using [[ I wouldn't have run into it at all, so I'm good either way with keeping or removing the partial matching