Format a string given a string template and some context.

and I had my head on backwards, that "$\{" doesn't work as advertised in the doc string - so that's an open question: "how do you represent the string "${"?

Replying to tschaub:

and I had my head on backwards, that "$\{" doesn't work as advertised in the doc string - so that's an open question: "how do you represent the string "${"?

I think that optimized performance is crucial to this function, so I would propose to represent "${" by "${${". I created a patch for this, which also handles the "}" character without a preceding "${" correctly. Performance is improved by starting the loop at 1 instead of 0 - the first token never needs to be replaced.

Improved version of the original patch. Allows to represent "${" by "${${" and handles "}" without preceding "${" correctly.

process a templated string

That's nice.

I made a few other changes. Ignore any items that start with "}" (last == 0). Also, variable assignment is cheap, array index is expensive - so I set item = tokens[i]. Now you do two array lookups instead of four on successful match (didn't test performance here).

Finally, I didn't get the bit about adding "}" on the end (line 122 in your patch). Please set me straight if I missed something. Better yet, add a new test case.

The tests are pretty simple, but I think they're enough to get this in - if someone reviews in IE (can't remember if undefined.toString() == "undefined" there too).

Replying to tschaub:

I made a few other changes. Ignore any items that start with "}" (last == 0). Also, variable assignment is cheap, array index is expensive - so I set item = tokens[i]. Now you do two array lookups instead of four on successful match (didn't test performance here).

Thanks for that information. I was not aware that array index access is expensive.

Finally, I didn't get the bit about adding "}" on the end (line 122 in your patch). Please set me straight if I missed something. Better yet, add a new test case.

The whole block you are referring to is never called, and to be honest I do not remember why I added it. I added two unchanged strings that I thought that would need this, but it was simply a mistake.

The tests are pretty simple, but I think they're enough to get this in - if someone reviews in IE (can't remember if undefined.toString() == "undefined" there too).

I ran the tests in IE, and they failed. It seems IE does not support "delete". So I had to move the undefined test above the window context test. Now the tests are all passed.

I attached a new patch with fixed tests. Everything is fine now.

Fixed tests, are now ok in FF, IE and Opera

Committed in r5317. Thanks!