On 2/09/2026 01:43, Bald Eagle wrote:
> So, I cannot see parts of this reply, presumably due to UTF-8/non-ASCII
> characters, and when I tried to use Thunderbird it said "connection refused."
> 
> Is this a website policy change, or something that I need to fix in my
> Thunderbird application?

If you are referring to text missing after the word "that" in line 3 of the post you
are replying to, it should not happen when viewing the NG via NNTP in Thunderbird
itself.

There was an issue viewing it in the web view, though.

The part that reads "That’s a really nice render of the compound of five tetrahedra
— thank you for sharing it." had two characters that were not being handled when the
message was being rendered as UTF-8. These are the chars 0x92 at offset 0x26 and 0x97
at offset 0x61.

00000000  22 53 68 61 79 22 20 3c  6e 6f 6d 61 69 6c 40 6e  |"Shay" <nomail@n|
00000010  6f 6d 61 69 6c 3e 20 77  72 6f 74 65 3a 0a 48 69  |omail> wrote:.Hi|
00000020  2c 0a 54 68 61 74 92 73  20 61 20 72 65 61 6c 6c  |,.That.s a reall|
00000030  79 20 6e 69 63 65 20 72  65 6e 64 65 72 20 6f 66  |y nice render of|
00000040  20 74 68 65 20 63 6f 6d  70 6f 75 6e 64 20 6f 66  | the compound of|
00000050  20 66 69 76 65 20 74 65  74 72 61 68 65 64 72 61  | five tetrahedra|
00000060  20 97 20 74 68 61 6e 6b  20 79 6f 75 20 66 6f 72  | . thank you for|

Both of these represent undefined codepoints in UTF-8 and the result of passing them
through the PHP htmlentities() function with the options that were in the code before
was to cause the text between the two points to vanish.

There's been a number of complaints about this behavior so I did spend some time today
trying to at least get it to not disappear the text, and it does appear to work now,
though the characters in question are replaced by default glyphs.

The proper fix is to convert each message to UTF-8 according to its declared encoding
(we do store this in the DB along with the message), but I don't have time to do that
right now so what's there will have to do. I have been working on the webview code
recently (RSS and sitemaps) and this matters for those, too, so it should get fixed in
the not too distant future.

FWIW see three attachments: one with the old behavior, one with the current, and one
with what it would look like with the codepage taken into account.

-- Chris

Post a reply to this message

Attachments:
Download 'before.png' (40 KB) Download 'after.png' (45 KB) Download 'ideal.png' (53 KB)

Preview of image 'before.png'

Preview of image 'after.png'

Preview of image 'ideal.png'