1168

submitted 2 weeks ago by homesweethomeMrL@lemmy.world to c/science_memes@mander.xyz

153 comments fedilink hide all child comments

you are viewing a single comment's thread
view the rest of the comments

[-] cypherpunks@lemmy.ml 100 points 2 weeks ago

For some reason this post caused me to search to see if there are types of birdseed which are not suitable for ducks. From a few minutes of reading, I'm relatively sure now that at least all common types of birdseed are fine for them. But, along the way, I found this LLM slop https://plantnative.org/can-ducks-eat-bird-seed.htm (URL rendered unclickable since it is spam) which strongly implies that ducks are not in fact birds:

"Yes, ducks do eat bird seed. Although bird seeds are specifically prepared to meet the nutrient demand of birds. But still, bird seed can be served as a snack to ducks in limited amounts."
"Bird seeds are loaded with nutrients and can be beneficial for ducks the way they are for birds."

[+] StrixUralensis@tarte.nuage-libre.fr -6 points 2 weeks ago

URL perfectly clickable from PieFed

PieFed W

[-] SomethingBurger@jlai.lu 51 points 2 weeks ago

URLs inside code fragments should not be clickable. This is a Markdown parsing error.

PieFed L

[-] StrixUralensis@tarte.nuage-libre.fr -1 points 2 weeks ago

Why shouldn't they be ?

[-] markz@suppo.fi 27 points 2 weeks ago

Code blocks are for displaying raw text, ignoring all formatting.

[-] mic_check_one_two@lemmy.dbzer0.com 16 points 1 week ago* (last edited 1 week ago)

Because it’s meant to be a block of code, not something that formats using MarkDown rules. The point of code blocks is that you can use special characters and have them rendered as plaintext, instead of actually working as special characters. For instance, here’s the classic Reddit shrug:
¯_(ツ)_/¯

Notice that the left arm is missing? That’s because the backslash and underscores are both special characters. The backslash cancels out the underscore, making the backslash disappear. Now here’s the same shrug formatted as code:

¯\_(ツ)_/¯

Notice that the left arm is visible, (or at least, it should be), because none of the characters were treated as special characters. If the left arm is invisible on your client, that is actually an error in parsing the MarkDown formatting.

Tangentially, if I wanted to format that shrug without the missing arms, it would actually take three backslashes, like this:

¯\\\_(ツ)_/¯

The first backslash cancels out the second, making the second visible. Then the third cancels out the underscore, making the underscores visible, if I only used two, the backslash would be visible, but the underscores would disappear and make the head italicized. But by using three, we get this:
¯\_(ツ)_/¯

Notice how I was able to actually show how I used all three backslashes, using the code block?

[-] cypherpunks@lemmy.ml 10 points 2 weeks ago* (last edited 2 weeks ago)

compare the rendering of the test output (which i also wrapped in backticks to tell the markdown rendering to render it as code) in my other comment on lemmy vs on piefed.

Essentially, it is not possible to reliably identify the URLs in code (in a language that is not known to either PyFedi or its markdown parser), because they sometimes are adjacent to otherwise-URL-valid characters which actually terminate the URL in whatever syntax is being used inside the code block. So, even though it is sometimes harmless to auto-linkify inside a code block, it is also often wrong and therefore should not be (and generally is not) done.

But also, code formatting should be available as a way to disable auto-linkification when a post or comment author wants to (as I did in my original comment).

[-] StrixUralensis@tarte.nuage-libre.fr 5 points 2 weeks ago

Oh ok, I see

Thank you for explaining !

[-] cypherpunks@lemmy.ml 2 points 2 weeks ago

hmm, i see my first edit to this comment got federated to piefed but subsequent edits did not; the version after subsequent edits can be seen here on the lemmy instance i'm posting from

[-] cypherpunks@lemmy.ml 17 points 2 weeks ago

URL perfectly clickable from PieFed

PieFed W

that is not a W... i surrounded the URL with backticks, which means it should be treated as code and not made into a link.

cc: PyFedi developer @rimu@piefed.social

I had a glance at the source and don't see a very quick fix, but I think the solution involves getting rid of the auto-linkification here and instead have that be done by the markdown library. This issue indicates that the markdown library PyFedi is using is capable of doing that.

Here is a test I wrote which I think should pass which does not currently :)

diff --git a/tests/test_markdown_to_html.py b/tests/test_markdown_to_html.py
index 329b19be..108276c5 100644
***
a/tests/test_markdown_to_html.py
+++ b/tests/test_markdown_to_html.py
@@ -37,6 +37,12 @@ class TestMarkdownToHtml(unittest.TestCase):
         result = markdown_to_html(markdown)
         self.assertTrue("<pre><code>code block" in result)
 
+    def test_code_block_link(self):
+        """Test code blocks formatting containing a link"""
+        markdown = "```\ncode block with link: https://example.com/ \n```"
+        result = markdown_to_html(markdown)
+        self.assertEqual("<pre><code>code block with link: https://example.com/ ", result)

it currently produces this failure:

====================================================== FAILURES =======================================================
_______________________________________ TestMarkdownToHtml.test_code_block_link _______________________________________

self = <tests.test_markdown_to_html.TestMarkdownToHtml testMethod=test_code_block_link>

    def test_code_block_link(self):
        """Test code blocks formatting containing a link"""
        markdown = "```\ncode block with link: https://example.com/ \n```"
        result = markdown_to_html(markdown)
>       self.assertEqual("<pre><code>code block with link: https://example.com/ ", result)
E       AssertionError: '<pre[24 chars]ink: https://example.com/ ' != '<pre[24 chars]ink: <a href="https://example.com/" rel="nofoll[61 chars]e>\n'
E       - <pre><code>code block with link: https://example.com/ 
E       + <pre><code>code block with link: <a href="https://example.com/" rel="nofollow ugc" target="_blank">https://example.com/</a> 
E       + </code></pre>
E       +

tests/test_markdown_to_html.py:44: AssertionError

HTH, and thanks for writing free software!

[-] rimu@piefed.social 6 points 1 week ago

That does help very much, thank you!

I've committed a fix.

[-] brbposting@sh.itjust.works 4 points 1 week ago

PieFed / @cypherpunks@lemmy.ml W

[-] cypherpunks@lemmy.ml 4 points 2 weeks ago

oops, i read my test closer and realized it will never pass as-is due to being cribbed from another test that was using in instead of assertEqual (and the expected string not containing the closing tags) but I trust it conveys what i mean to