David Lutterkort

Yeah, that's unfortunate; we should just merge this as is, and live with the fact that we could have offered a better breakdown of the file.

The reason that `/` is not allowed in a key is that it is used to separate path components: if you allowed it, it would be impossible to know whether...

Off the top of my hat, changing this behavior would require the following: * On [this line](https://github.com/hercules-team/augeas/blob/master/src/lens.c#L575) use a regexp that only matches unescaped `/` * On [this line](https://github.com/hercules-team/augeas/blob/master/src/lens.c#L575), similarly,...

Yes, you are absolutely right. There is no need for the typechecker to worry about embedded `/` - when we match path expressions, we match them against trees. The `/`...

Hi Adam, first off, this is fantastic - getting Augeas to work on Windows has been a longstanding thorny issue, and I didn't know about the cygport effort (see for...

I can reproduce that, too. I am somewhat hesitant to touch this, since the Shellvars lens has grown to be quite complex (it parses a lot of actual shell scripts,...

I _think_ what is happening here is that because a comment 'in the middle' is removed, all other comments slide up one in their formatting, i.e., when `#comment[3]` is printed,...

@jreidinger if you want to play with that PR a bit that would be most appreciated. The big question in my mind is whether we should just change the behavior...

@mikhirev the surprise doesn't come from the fact that the file is parsed twice, but from how entries in the tree are associated with the stuff that gets ignored during...

@jreidinger Generalizing your approach to work with the full Augeas API would be really hard: in general, users can mix querying the tree and modifying it freely, so that if...