tags that contain a # symbol don't work right

[dkg]

there are three tags that appear to contain a # symbol in the current database:

• \\#AABILL
• \\#DEFENDENCRYPTION
• \\#SECUREAUSTRALIA

clicking on any of these links from the /tags/ URL leads to link targets like https://the-encryption-compendium.github.io/tags/%5C%5C#aabill/ -- this is problematic for at least three different ways:

• i suspect the backslashes are not supposed to be doubled -- indeed, they show up as single backslashes in the "Tags" entries that appear in a search for "reputation", though they show up doubled when you click through
• i suspect the backslashes are not supposed to be visible at all, that they somehow are meant to escape the octothorpe -- but that's not working.
• When you do click on one of the tags, the # gets treated as the URL's "fragment" delimiter so it doesn't land on the tag page at all.

The actual tag landing page appears to have the double-backslash (represented as %5c) in it, followed by the percent-escaped octothorpe (as %23): https://encryptioncompendium.org/tags/%5c%5c%23aabill/

Fixing this correctly seems likely to be fiddly -- there are likely to be several places that need fixing. But perhaps the simplest way to avoid the problem is to forbid the use of the # character in a tag label. This kind of constraint might be enforced as part of the bibtex consistency check that should probably be run before a site build.

[kernelmethod]

Yeah, this is part of a bigger bug that I flagged in #16, although the # symbol is a particularly obnoxious case of that bug. As you mentioned, fixing this bug is difficult, especially when you look at some of the other cases that can occur (tags with spaces or unicode in them, tags like 9/11 with / in them, etc.). My go-to solution for this would be to try to slugify the tags as I did with the compendium entry titles, but Hugo makes it difficult to do that kind of thing.

I'm thinking it might be better to work around all of this by just getting rid of the /tags/ URL altogether for looking up articles with a given tag, since I don't really think it meshes very well with the rest of the site anyways. IMO it'd be better to change the search page to allow lookups based on tags -- for example, include a little drop-down box where folks can select a handful of tags that they want to search on, or something to that effect.

[dkg]

fwiw, I like having the /tags/ URL -- it'd be even neater with a "tag cloud" that emphasizes popular tags based on their frequency. But most of the problems outlined above wouldn't be solved by removing it (e.g. each tag's own page, the tag-specific links that show up on each entry's page). Why not just constrain tags directly to alphanumeric ASCII plus some non-objectionable punctuation (e.g. . and - and _) until these other problems are solved?

As long as the tags are all entered and maintained by staff, sticking to those constraints shouldn't be problematic.