Re: join strips backslash - feature or bug?

Larry W. Virden wrote:
On May 3, 9:42 pm, "slebet...@xxxxxxxxx" <slebet...@xxxxxxxxx> wrote:
Consider that different implementations of Tcl are free to quote the
string in different ways as long as it generates a valid list. Also
consider that future and past versions of the "official" Tcl are also
free to quote the string differently from the current 8.4/8.5. This
means that the most "robust" way to document the behavior of [list] is
that it will quote "as necessary".

In light of this, IF it is documented, then it should not be
documented in the actual man page but in the implementation's README/
changelog/errata instead.

In my opinion, this is an argument FOR it to be documented in the man
page. The README, etc. are not always available. And so details on how
a command in a language works should always appear in the reference
documentation for that language.


No, it is an internal detail that should not be relied on in any way by
end users, use list commands to build and access list and the exact form
of quoting is irrelevant. Same reason that individual man pages for tcl
script level commands have no details about byte codes used, or internal
object reps - these are implementation details that can change and are
(and should be) opaque to the end users (script writers). The only place
it needs to be worried about is at the maintiner of the commands that
handle it, so the documentation should be good comments and or change
logs if it is modified. (Or other maintainer used documentation).

Bruce

PS - this is just *my* opinion, I am not a member of the TCT or a maintainer of any part of tcl - just a long time user of the language.
.

Relevant Pages