-
• #7602
Thank you @gbj_tester and @Velocio!
-
• #7603
So I went and tested... and yup, even things like HTML
meter
works:at 20/100 at 50/100 at 80/100 Cannot imagine anyone using it... but the gist is, if it's not malicious then it was enabled.
-
• #7604
Hyperlinking to a YouTube video while removing 'https://' (in order to prevent the embedding of the video) results in some funny business:
leads to:
https://www.lfgss.com/conversations/127109/
Code:
[Clicking here](youtu.be/watch?v=7oS7qfQ5Z9k&t=18s)
Hovering over the 'Clicking here...' link above shows this link:
-
• #7605
that is weird
traced it back to Django and Python... it doesn't seem to understand not to read beyond the 131364, and somehow Python manages to extract 127109 from the resulting stringification mess, and then that just happens to actually exist as a thread, and so the redirection works.
beats me... undocumented behaviour is still not a security risk as all permissions and existence checking is applied, so I'll happily ignore it.
relative links though... those aren't going to work
-
• #7606
Interesting. Anyway, I’m able to circumvent this behavior by replacing ‘youtu.be’ with ‘https://m.youtube.com’.
It won’t embed the video, and the link works.
Code:
[the link works](https://m.youtube.com/watch?v=7oS7qfQ5Z9k&t=18s)
-
• #7607
Why do you not want to embed?
-
• #7609
Rephrase 🤷 you already know the link will embed.
-
• #7610
But I don’t want to. ;)
-
• #7611
Post it, copy the generated short link, edit it.
The generated short link for the above is https://microcosm.app/out/T5jSi
-
• #7612
That also works! Nice.
I’m able to circumvent this behavior by replacing ‘youtu.be’ with ‘https://m.youtube.com’.
-
• #7613
Well that's a bug, so I'll fix that.
-
• #7614
😭
-
• #7615
when you hover over a blue external link on a post, the bottom left of my screen shows the link that you are attempting to link to. it always shows https://microcosm......... etc
is there anyway you could actually show the website it is linking to so people can avoid going onto websites like the daily mail / the sun, or other websites that they wish to avoid. it is sometimes difficult to tell what website you are visiting before you click the link. -
• #7616
yeah, I should change that.
it has been very useful at blocking malicious websites, cryptojackers, and spam... but it's not like I use it a lot (as not a lot is reported).
the original intent was to redirect via a point I control so that it was possible for me to moderate where things went to.
a side benefit is that I'd have visibility over things like links through to affiliates and could hold those third parties to account.
I've never used the analytics aspect of the redirect, but had used the blocking aspect... but arguably it's no longer needed.
however to fix it now requires major surgery, and I'm kinda waiting for me to rewrite the frontend... but I've not made much progress as my day job consumes my time and effort.
it's on the list of features to remove, it's a failed experiment in safety + accountability, it does work, but probably isn't worth the inconvenience created.
-
• #7617
thanks for the explanation. for us non i.t. types we just think it's a simple thing but have no idea about all the things that go on behind the scenes.
it's especially confusing when someone creates a link but then types a description of what the link is instead of showing the https .... for example " cheap gumwall tyres " so you have no idea if it's a mark all read type scenario, daily heils website or an actual link to cheap gumwall tyres.
if it goes through you and you moderate that must help stop people linking malicious websites or websites with known viruses on via databases of dodgy websites and ones to avoid. but it's rare i click on a website that i don't know or has some weird web address just in case of scams viruses
-
• #7618
Not sure if correct place to ask, but is there a reason the top threads I see in General aren’t in order of newest replies?
1 Attachment
-
• #7619
Yes... more easily seen on tablet or desktop view.
4 of those are pinned to the top (and sorted by last reply).
3 of those are themselves forums (and sorted by magic - i.e. some trade-off on number of comments, items, recency).
and then everything else is sorted by last reply time.
1 Attachment
-
• #7620
Ahhhh got it. Thanks
-
• #7621
Has the DNS been been hijacked? I'm seeing this:
1 Attachment
-
• #7622
the old microco.sm domain expired, and it appears somewhere in there it affected the Cloudflare CNAME flattening when it got nuked.
I've removed microco.sm from my Cloudflare account, which prevents the CNAME flattening... and that resolved it.
-
• #7623
I was very fucking worried briefly.
-
• #7624
This.
I even joined the fb group to see the end times had arrived.
-
• #7625
Worst morning of my life
IF IN DOUBT, USE MARKDOWN: https://www.markdownguide.org/cheat-sheet/
You have three types of formatting available:
Those are applied in that order to produce HTML (we convert the BBCode to Markdown, then convert the Markdown to HTML), and the final output is then sanitised to remove HTML that can be used maliciously (i.e. to obtain someone else's cookies or session).
The BBCode we support:
The final accepted list of HTML is captured within the user generated content policy, where anything not matching the following HTML is removed from the post:
Additionally we add
nofollow
as arel
on all fully qualified links, and we will also add_blank
as atarget
on fully qualified links. HTML entities are also escaped, the output is pure HTML and nothing ambiguous will remain.Finally, any comment entirely made of capital letters (according to a lack of
unicode.IsLower()
rune being present) will be run (in it's totality) throughunicode.ToLower()
... which means anyoneSHOUTING
will have their post converted toshouting
.Essentially... you almost do not need to worry about it, we support multiple types of formatting and way more than you imagine, and it's very comprehensive so even if you're cutting and pasting in rich content from elsewhere it mostly works.
Notable omissions though, we do not support anything not explicitly mentioned above, but specifically we have not mentioned:
But we will recognise links to some places (like YouTube) and convert those into embeds for you. We cannot take their HTML as input as it would mean disabling security features, but we can produce their HTML from a link.