aboutsummaryrefslogtreecommitdiff
path: root/doc/bugs/escaped_style_tag_becomes_elyts.mdwn
diff options
context:
space:
mode:
authorsmcv <smcv@web>2014-10-05 08:40:27 -0400
committeradmin <admin@branchable.com>2014-10-05 08:40:27 -0400
commitaf94275af1c168bd51a620d9998ead93892cfd5d (patch)
tree43735d0a5bc3ddbd81baeda600bdd88ff7056862 /doc/bugs/escaped_style_tag_becomes_elyts.mdwn
parent4a90553d7f34d7178beb96489a5b39a545c532ac (diff)
downloadikiwiki-af94275af1c168bd51a620d9998ead93892cfd5d.tar
ikiwiki-af94275af1c168bd51a620d9998ead93892cfd5d.tar.gz
new bug report
Diffstat (limited to 'doc/bugs/escaped_style_tag_becomes_elyts.mdwn')
-rw-r--r--doc/bugs/escaped_style_tag_becomes_elyts.mdwn28
1 files changed, 28 insertions, 0 deletions
diff --git a/doc/bugs/escaped_style_tag_becomes_elyts.mdwn b/doc/bugs/escaped_style_tag_becomes_elyts.mdwn
new file mode 100644
index 000000000..24d8fbf7e
--- /dev/null
+++ b/doc/bugs/escaped_style_tag_becomes_elyts.mdwn
@@ -0,0 +1,28 @@
+When IkiWiki uses discount to implement [[plugins/mdwn]] rendering,
+there is a workaround for <https://rt.cpan.org/Ticket/Display.html?id=74016>:
+
+<pre><code>$t=~s/&lt;style/&lt;elyts/ig;
+my $r=Text::Markdown::Discount::markdown($t);
+$r=~s/&lt;elyts/&lt;style/ig;
+</code></pre>
+
+However, this workaround also applies to indented text or text in backticks:
+if you write <code>there is a bug involving the `&lt;style&gt;` tag</code>,
+or use indentation like
+
+<pre><code>you can use this markup:
+
+ &lt;style type="text/css"&gt;...&lt;/style&gt;
+</code></pre>
+
+then that gets turned into `&lt;elyts` in the HTML too. This makes it quite
+difficult to talk about HTML on an IkiWiki instance (I had to use raw HTML in
+this bug report's source to avoid the bug).
+
+I think the side-effect of the workaround is more damaging than the actual bug
+being worked around: I've never wanted to write inline style tags in the body of
+a Markdown page (which isn't even valid HTML) but I have certainly wanted to
+discuss style markup several times. The first couple of times I saw this happen,
+I thought it was some sort of misguided anti-cross-site-scripting filter...
+
+--[[smcv]]