# 2024-02-09 Oddµ namespaces and live updating > It's a slow morning. I had my two cups of coffee, ate my pancakes with plum jam, spent time online reading Mastodon, talked to my wife about wannabe-axe murderers on Swiss trains, looked up stuff about taxing the rich… => https://www.tax-the-rich.eu/ taxing the rich I've been working on Oddmu again. My goal is to be able to reproduce the Campaign Wiki setup: => https://campaignwiki.org/ Campaign Wiki * people can get a "space" for their site by signing up in some way * they can make changes to the look and feel of their site (change the headers, footers, colours, i.e. make changes to both the HTML template and the CSS) * feeds, list of changes and all that are limited to their site Using Oddmu would also result in the following: * sites are no longer publically editable (which isn't a requirement for such spaces like role-playing game campaigns) Oddmu now has the following features: * when editing pages in a subdirectory, the `index` and `changes` page in the same directory are updated (instead of the root directory) * when HTML templates exist in the same directory, those are used instead of the templates in the root directory (e.g. Oddµ) * when uploading a HTML template, it is loaded immediately (no more server restart) => oddmu/index Oddµ Once I had this implemented using a library that watches the filesystem for changes, I realized that I could use the same feature to index new pages. This brings me closer to having online and offline writing feature equivalence: Using the web site to create a page should be equivalent to writing a page elsewhere and uploading it to the site via rsync. (Offline you can use the `oddmu notify` command to add links to `index`, `changes` and hashtag pages.) I think I'm getting close! ​#Oddµ Also, code complexity is through the roof, now. All the global maps aren't thread-safe so they need mutexes. There's a weird Heisenbug I have where I run the test suite and then `view.html` is gone. The next test run then fails, of course. Thinking it was related to the order of tests, I started running `go test -shuffle on` and found a plethora of *other* bugs. And I've seen at least two occasions where I noticed the disappearing `view.html` bug, noted the shuffle id, reran the test with that id, got the error again, tried for a third time, and then tests passed. So… it must be a race condition of some sort. How aggravating. Maybe something got lost, with that last set of changes. I do feel, however, that I've found a lot of bugs… Just now, for example, to help prevent Oddmu from showing any directories or files with a path segment starting with a period… **2024-02-14**. I need to think about an archive handler that allows you to get a zip file for the current directory. (I guess zip files are more accessible than .tar.gz files? Like, for Windows and macOS…) And the archive needs to take ODDMU_FILTER into account! => 2024-02-14-oddmu-bug ODDMU_FILTER Done. ✔️