Closed gaborcsardi closed 3 years ago
@rundel Thanks for looking at this. I think it is definitely an improvement and hopefully we were conservative enough not to break anyone's code.
There is still this to fix:
se = gh::gh(
"GET /search/code?q=foobar",
.limit=101
)
lengths(se)
#> total_count incomplete_results items
#> 2 2 200
se$total_count
#> [1] 18043087 18080930
se$incomplete_results
#> [1] TRUE FALSE
In the future I am hoping that we can use r-lib/webfakes and the info from the spec to write a test suite with reasonable coverage, and then we can be more confident about handling the output of the API better.
Oh, yeah the issue with that code is this line:
if (is.atomic(z)) unique(z)
and of course the API provided different values for total_count
and incomplete_results
for the two responses. I'll tweak that a bit and will use the latest value for these two fields.
Use 'total_count' and info from the schema on how to calculate the "length" of a response.
Fixes #135.