Re: [PATCH v2.5 1/5] libxenguest: support zstd compressed kernels

[Ian Jackson <iwj@xxxxxxxxxxxxxx>]

Jan Beulich writes ("Re: [PATCH v2.5 1/5] libxenguest: support zstd compressed 
kernels"):
> Anyway - I guess we should continue from v3, which I hope to post
> later this morning.

Thanks, yes.  I just wanted to reply to one bit of this:

>   My initial attempt to avoid configure failing was to specify [] as
> the 4th argument. This, to me, would have felt the half-way natural
> indication that I don't mean anything to be done in the failure
> case, neither autoconf's default nor anything else. [true], otoh,
> already feels like a workaround for some shortcoming.

configure.ac is m4 input text.  In m4, missing arguments to a macro
are experienced within the macro definition almost identically to an
empty argument.  (There are some GNU-m4-specific facilities that make
it possible for the macro to tell the difference, but it would be
highly unidiomatic to do that for an autoconf macro like this one.)

So naturally [] would do the same as a missing argument.  It's
possible that [ ] would have worked as well as [true] (depending on
the levels of quoting etc.), but [true] is more idiomatic to me.  It's
the usual way of spelling an explicit no-op in shell.

Ian.