Build error: Missing value in while (tries < max_tries && Sys.time() < deadline (httr2)

[jhollway]

I am encountering the same issue when trying to build a pkgdown site for several packages. Essentially the site building process aborts when trying to build the news (it also occurs when I use pkgdown::build_news()).

pkgdown:::build_site()
...
── Building news ───────────────────────────────────────────────────────────────
Error: 
! in callr subprocess.
Caused by error in `while (tries < max_tries && Sys.time() < deadline) { …`:
! missing value where TRUE/FALSE needed
ℹ See `$stdout` and `$stderr` for standard output and error.

From the backtrace(below), it seems to be an issue what what pkgdown:::pkg_timeline() is passing to httr2::req_perform(), as that is where the error is thrown.

1. pkgdown:::build_site()
2. pkgdown:::build_site_external(pkg = pkg, examples = examples, run_dont_run = run_dont_run, …
3. callr::r(function(..., cli_colors, hyperlinks, pkgdown_internet) { …
4. callr:::get_result(output = out, options)
5. callr:::throw(callr_remote_error(remerr, output), parent = fix_msg(remerr[[3]]))
---
Subprocess backtrace:
1. pkgdown::build_site(...)
2. pkgdown:::build_site_local(pkg = pkg, examples = examples, run_dont_run = run_dont_run, …
3. pkgdown::build_news(pkg, override = override, preview = FALSE)
4. pkgdown:::build_news_single(pkg)
5. pkgdown:::data_news(pkg)
6. pkgdown:::pkg_timeline(pkg$package)
7. httr2::req_perform(req)
8. base::.handleSimpleError(function (e) …
9. global h(simpleError(msg, call))

I should add that this occurs for packages that are on CRAN and that have (had) pkgdown sites built in the past without issue.

[jayhesselberth]

I can't reproduce this locally, might be an intermittent connectivity issue. Please re-open if you continue to see this.

[jhollway]

This error continues, which @jaeltan can reproduce locally. I don't have permission to reopen the issue, however.

[jhollway]

We are encountering this issue for example with messydates

[jayhesselberth]

OK, I can reproduce this. But I'm not exactly sure what's going on. It seems like a possible issue in httr2. Maybe @hadley has some insight.

Edit: now after going back to this, I can't reproduce the error below. Perhaps this actually is an intermittent internet issue?

If I run the following in the console (part of pkgdown:::pkg_timeline()):

package <- "pkgdown"

url <- paste0("https://crandb.r-pkg.org/", package, "/all")
req <- httr2::request(url)
req <- httr2::req_retry(req, max_tries = 3)
req <- httr2::req_error(req, function(resp) FALSE)
resp <- httr2::req_perform(req)

I see this error, same as above:

Error in `while (tries < max_tries && Sys.time() < deadline) {
    retry_check_breaker(req, tries, error_call = error_call)
    sys_sleep(delay, "for retry backoff")
    n <- n + 1
    resp <- req_perform1(req, path = path, handle = handle)
    req_completed(req_prep)
    if (retry_is_transient(req, resp)) {
      tries <- tries + 1
      delay <- retry_after(req, resp, tries)
      signal(class = "httr2_retry", tries = tries, delay = delay)
    } else if (!reauthed && resp_is_invalid_oauth_token(req, resp)) {
      reauthed <- TRUE
      req_auth_clear_cache(req)
      req_prep <- req_prepare(req)
      handle <- req_handle(req_prep)
      delay <- 0
    } else {
      break
    }
  }`:
! missing value where TRUE/FALSE needed

BUT if I run the same code as a reprex, it's fine:

package <- "pkgdown"

url <- paste0("https://crandb.r-pkg.org/", package, "/all")
req <- httr2::request(url)
req <- httr2::req_retry(req, max_tries = 3)
req <- httr2::req_error(req, function(resp) FALSE)
resp <- httr2::req_perform(req)

resp
#> <httr2_response>
#> GET https://crandb.r-pkg.org/pkgdown/all
#> Status: 200 OK
#> Content-Type: application/json
#> Body: In memory (49723 bytes)

^{Created on 2025-02-27 with reprex v2.1.1}

[hadley]

I can't figure out why that line would even fail. As far as I can tell from reading the source code, none of tries, max_tries, deadline or Sys.time() can ever be NA.

There are only two lines that modify tries:

tries <- 0
tries <- tries + 1

There's only one line that sets max_tries

max_tries <- retry_max_tries(req)

That will always be 3 in this case.

There's only one line that sets deadline:

deadline <- Sys.time() + retry_max_seconds(req)

And here retry_max_seconds() will be Inf.

So unless my reasoning above is wrong (and if it was I think we'd see this problem a lot), then the mostly likely explanation is the Sys.time() is returning NA? But that seems extremely unlikely.