Clarification regarding what characters may precede tags

[chris48s]

Fairly succinct one:

[ ] This item has a #tag
[ ] This item has a#tag

Does the second line have a tag?

[jotaen]

Currently, yes.

The reasoning is to allow tags to be recognised in use-cases like these:

[ ] Do something (#tag1, #tag2)
[ ] Do other thing #tag1/#tag2

Initially, I tried to distinguish that more fine-granularly, and to allow tags to be surrounded/preceded only by punctuation characters. I ended up abandoning that, though, because I couldn’t come up with rules that are reasonably straightforward (e.g. can be expressed via a “simple” regex) and universally applicable (in the sense that they work in all languages, not just in English or other western languages). For reference, this was the original ticket and this the corresponding spec change.

The current situation has two downsides: (a) the rules are not 100% clear currently, so your original question is hard to answer by merely looking at the spec; and (b), it creates false positives / ambiguities, such as:

• [ ] Worked on website: https://example.org/foo/#some-anchor → Recognises #some-anchor as tag
• [ ] Change my master password to be bN144#y6Q!Jc → Recognises #y6Q as tag

So far, this didn’t seem to have caused too much trouble in practice. I’m nevertheless open to improving it.

[chris48s]

That makes sense. The URL anchor false positive is particularly annoying, but I don't think there is a nice way to recognise #tag1/#tag2 but not https://example.com/#anchor and keep the spec/parsers simple. Without the ability to escape a hash character (not necessarily desirable), I don't think there is any way to completely eliminate false positives.

The current situation has two downsides: (a) the rules are not 100% clear currently, so your original question is hard to answer by merely looking at the spec

Yes. This is the reason why I asked the question :) The spec should make it clear. Perhaps include an example to clarify it?