[racket-dev] [plt] Push #22553: master branch updated

From: Robby Findler (robby at eecs.northwestern.edu)
Date: Tue Apr 26 13:19:08 EDT 2011

On Tue, Apr 26, 2011 at 12:09 PM, Stevie Strickland
<sstrickl at racket-lang.org> wrote:
> On Apr 26, 2011, at 1:05 PM, Robby Findler wrote:
>> On Tue, Apr 26, 2011 at 12:01 PM,  <sstrickl at racket-lang.org> wrote:
>>> f5de8bd Stevie Strickland <sstrickl at racket-lang.org> 2011-04-26 12:57
>>> :
>>> | Move scmxlated source for slatex into private.
>>> |
>>> | Anyone using the sole export from slatex.rkt should really be using the
>>> | functions provided by slatex-wrapper.rkt instead, which I imagine is why
>>> | this has never been documented.
>>> :
>>>  R collects/slatex/{ => private}/slatex.rkt (100%)
>>>  M collects/slatex/README                       |    4 ++--
>>>  M collects/slatex/slatex-wrapper.rkt           |    2 +-
>>>  M collects/slatex/tests/test-docs-complete.rkt |    1 -
>> Any reason not to just document it instead? This code has been
>> unchanging forever and I don't think we want to break old scripts
>> whimsically.
> I would imagine that any old scripts would be using the documented slatex-wrapper interface, especially since there's a lot of setup work done in that interface that would have to be duplicated otherwise.  Do we have reason to believe that people have been using the undocumented `slatex::process-main-tex-file' function instead?

I don't know of people who are, but I think that in situations like
these we should err on the side of not-breaking-code.

In this case, I would document this function as "roughly like the
@racketmodname[slatex] library function X, provided here for backwards
compatibility. New code should use X." or similar.

Is that any more difficult than what's been done?


Posted on the dev mailing list.