| japhb | Oh that's annoying. I went through it with both `ftp` and `curl`, and with both the old (/Public/UCD/latest/ucd/) and new (/Public/UNIDATA/) paths, and none of them work directly. | 02:54 | |
| However it appears that an FTP `ls` command at least *shows* the link destinations, so perhaps it's a matter of parsing the file listing. Here's what I see when I just curl against the directory containing the symlink: | 03:02 | ||
| $ curl --ftp-method nocwd -s ftp://ftp.unicode.org/Public/UCD/ | |||
| lrwxrwxrwx 1 ftp ftp 41 Sep 16 15:16 latest -> /Public/18.0.0 | |||
| Then replacing UCD with 18.0.0 gives a full listing. | 03:03 | ||
| So it probably just needs a new routine to parse out the symlink destination and patch the URL with it | |||
| ShimmerFairy | Using Filezilla I saw 550 permission errors, so I assume the server just set up the symlinks wrong. What I decided to do was swap over to HTTPS since that would be easier than swapping to a more low-level use of FTP. | 03:39 | |
| I also get the impression that perhaps the FTP server isn't really intended for use anymore, given you'd be hard-pressed to find any mention of it on the Unicode site (I could only find a reference in a fossilized page from 2005). You also can't establish a secure connection to it, which enhances that "old and unmaintained" feel. | 03:41 | ||
| Voldenet | but you don't need all files I think: github.com/MoarVM/MoarVM/blob/main...nload.raku | 06:32 | |
| if I'm reading that right and picking right file, it could simply use `unicode.org/Public/` instead of `ftp://ftp.unicode.org/Public` | 06:34 | ||
| eh, right, nevermind MAPPINGS are a folder, not a file | 06:35 | ||
| ah, but if you look closely, it just fetches 2 files from mappings | 06:44 | ||
| just using `"unicode.org/Public";` mostly works, and if you need the latest files, `wget -np -r` would work | 06:50 | ||
| (due to structure changes etc.) | |||
|
08:02
vrurg_ joined
08:04
vrurg left
|
|||
| ShimmerFairy | I need to drag myself to bed, but I need to mention something first. Thanks to a new test case Unicode added to their normalization tests, I uncovered an ancient bug in how MoarVM composes hangul characters. (Which I will of course share a fix for once I get everything sorted out.) | 13:06 | |
| m: my $test = Uni.new(0xAC00, 0x11A7); say $test».chr; say $test.Str; # the first say demonstrates what you should be seeing in the second; the second drops a codepoint. | 13:07 | ||
| camelia | (가 ᆧ) 가 |
||
| lizmat | ++ShimmerFairy++ | 13:08 | |
| ShimmerFairy | Oh, and I spotted another similar bug in the same part of the code. (In short, it's using ≤ or ≥ in places where < or > is intended.) | 13:10 | |
| But like I said, I need to sleep first, then I can better juggle all the edits in all the repos I've been bouncing around today. | 13:11 | ||
| lizmat | cool thing to look forward to! | 13:12 | |
|
14:51
[Coke]_ joined
14:53
[Coke] left
15:44
[Coke]_ is now known as [Coke]
|
|||
| japhb | ShimmerFairy++ # Finding all the bugs | 16:36 | |