I've been going back between the following two entries for keyword-tab search engines from the address bar (e.g. "codesearch<Tab>search phrase<Return>").

https://codesearch.wmflabs.org/search/?q=%s
https://codesearch.wmflabs.org/search/?q=%S

The difference between the use of lowercase or uppercase S. The difference between these is that lowercase S uses percent-based url-encoding for the phrase, and uppercase S substitutes the phrase directly in the url.

Both of these have issues for me:

With lowercase S, a query is "foo=&" encoded as /search/?q=foo%3D%26 which is desirable and runs the query the same way I entered it before encoding. The problem is that use of a space (e.g. foo bar) results in /search/?q=foo+bar (using Chrome, latest stable), which is interpreted by Code Search as a literal plus sign instead of a space.

I don't know if there is a way to force the browser to use %20 for space encoding, as that would solve this problem.

It can be worked around by using uppercase S, which will pass the space as a literal part of the query, but then any special characters like ampersand and equals will be parsed as part of the url query string, instead of the value of a specific query parameter, and thus effectively drops part of the string. E.g. foo=&bar results in`/search/?q=foo=&bar` which surprisingly doesn't search for anything (why?), and foo&bar results (naturally) in a search just for foo as the ampersand terminates the value for q.