Anchor macros randomly break.

[Eric Riddle]

Hi. We're in the process of inserting anchors within our documentation and we're finding that the anchors are randomly breaking and renamed.

For example, we create an achor called Requirements

Later, when testing, we find that links to that anchor have been renamed to Requi. We fixed it, and later we found it called Requ.

Of course, any link within the document that links to #Requirements is now broken.

Any ideas?

Edit: Update with clarification:

The anchor names remain the same. It's the links that get renamed.

For example, I might have text that says "check the requirements section" and the link is to #requirements.

Later I check and the text is the same, but the link is now #requi and doesn't jump to the anchor.

[Eric Riddle]

Matthew, I'm sorry I wasn't clear.

The anchor names remain the same. It's the links that get renamed.

For example, I might have text that says "check the requirements section" and the link is to #requirements.

Later I check and the text is the same, but the link is now #requi and doesn't jump to the anchor.

[Matthew J. Horn]

When you say it has been renamed, do you mean the actual <a> (anchor) tag within the page has been "renamed" or the text in the link to that anchor tag has been "renamed".

[Matthew J. Horn]

Huh, I use a lot of local anchors and haven't seen this before. I can't think of why it might be happening. Can you reproduce this, or is it fairly random? I'd say if it's reproducible, you should file a bug. If not, hopefully, someone else can chime in and suggest what might be happening.

[Eric Riddle]

It's fairly random. Me (and my documentation team) just noticed it this morning. I haven't found a way to reliably reproduce it.

Our workflow right now is focused on adding links, anchors, and cross-referencing articles. So we're going through articles and adding this, then we check each others' work. We fixed links, but now we're finding links that we /know/ we fixed broken again.

[Matthew J. Horn]

I guess one thing I would do to diagnose would be to look at the history of the page and diff the page with the previous to see when the change occurs. Not sure if this is going to lead to anything, but at least there's a chance that a pattern will emerge.