29 Aug 2025 |
fzakaria | i learned about some hack that the compiler does for rvalues that are also created as function variables to make their lifetime expended. | 23:43:01 |
fzakaria | expanded* | 23:43:13 |
fzakaria | i see this | 23:43:32 |
fzakaria |
Clang-Tidy: Std::basic_string_view outlives its value
| 23:43:33 |
fzakaria |
Clangd: Object backing the pointer 'modifiedUrl' will be destroyed at the end of the full-expression
| 23:43:43 |
John Ericson | ah yeah | 23:44:04 |
John Ericson | I only did one of the two path prefix thinggies | 23:44:12 |
John Ericson | I didn't see the other | 23:44:14 |
fzakaria | why did you chang eit to push back empty string? | 23:44:50 |
fzakaria |
path.emplace_back("");
| 23:44:52 |
fzakaria | (it was pushing back forward slash before) | 23:45:14 |
John Ericson | give it a pull | 23:48:56 |
fzakaria | oh i was gonan land them ;P | 23:49:11 |
John Ericson | oh sorry | 23:49:22 |
John Ericson | I did that because the initial "" makes an initial / | 23:49:33 |
John Ericson | I get a test failure now | 23:49:56 |
John Ericson | but at least local and CI should agree | 23:50:00 |
fzakaria | from what ? | 23:51:22 |
fzakaria | i thin kits from the new try | 23:52:39 |
fzakaria | parseURL actually parses scp like... | 23:52:50 |
fzakaria | that's why i had the if guard | 23:52:54 |
fzakaria | (it parses it wrong... but it succeeds the parse) | 23:53:07 |
fzakaria | i.e. github.com:/owner/repo.git | 23:53:13 |
fzakaria | (probably another boost bug to open) | 23:53:18 |
fzakaria | k i added those guards back and pushed. | 23:55:38 |
fzakaria | if (auto r = boost::urls::parse_uri(url); r && r->has_scheme() && r->has_authority()) {
return parseURL(url);
}
if (hasPrefix(url, "file:"))
return parseURL(url);
| 23:55:45 |
30 Aug 2025 |
John Ericson | we should not use two UIR parsers | 00:01:46 |
John Ericson | if we need to check schem and authority | 00:01:55 |
John Ericson | we can do it on our parseURL | 00:01:59 |
fzakaria | gonna make dinner; put a comment on the PR | 00:02:50 |