Bruno Le Floch

Now there is partial support with `\unravel_get:nnN { mute } { "some code" } \l_result_tl`. However since I had not planned this feature from the start I almost certainly abused...

It would be good to let the user customize things a bit in this area. Currently, the internal `\g__unravel_speedup_macros_bool` could give some control.

Besides error recovery, one advantage of having a better handle on macros is that it opens up the hope to one day support catcode changes. The idea would be to...

Adding some logging to `\@@_generate_internal_variant:n`, I find the following list of exp args that do not appear at all in `l3expan`. I did not check whether the documentation and implementation...

Setting the (internal) "speedup macros" boolean to false fixes the problem. One issue is that `\@@_macro_call_quick_loop:NNN` calls `\@@_input_gpop_item:NF` and loses track of whether the argument was braced or not. Otherwise...

Actually the problem is more dire than I had understood. With `\def\test#1{}`, expanding `\test{{{{{\par}}}}}` leads to a Runaway error and leaves five closing braces in the input stream. The only...

There should be different kinds of steps, most obviously steps which perform a non-expandable primitive and steps which simply perform expansion. These should be clearly differentiated. It may be useful...

This is related to #17.

That's a bug, but difficult to fix as the expansion-preventing aspects of `\noexpand` are very temporary and a true "`\notexpanded:\foo`" token cannot be stored. Instead I have cheated by introducing...

Thanks. I've logged adding XeTeX's `\strcmp` as a separate issue, #23. This one is more about testing that all primitives that I claim to support are indeed supported. (I fear...