>>>>> On Sun, 4 Mar 2007 15:47:09 -0800, Eric Wilhelm <[EMAIL PROTECTED]> said:
> # from Andreas J. Koenig
> # on Sunday 04 March 2007 12:25 pm:
>> > Ok, pretend I said "> /tmp/foo" or "| thbbt" in the example. In
>> all > three cases, it should not be prompting without a non-terminal
>> STDIN.
>>
>> I'm sorry, I cannot agree. If people deliberately keep their STDIN on
>> the terminal, then the terminal should be treated as the legitimate
>> owner of the input stream. That is exactly the situation for which I
>> filed this bug.
> You're saying you want the prompt to be printed when STDOUT is not a
> terminal (possibly even /dev/null) and you're prepared to answer
> *interactively* on STDIN at the terminal?
Yes.
> What's the use-case for this?
Capture output.
> Some sort of half-harness? Looking at your original post, I would not
> be expecting you to want a prompt.
Yes, sorry, the beginning of this bugreport was very vague because it
was cumbersome for me. This was because I did not find out what was
going on at all. If a terminal hangs for no apparent reason, does not
hang in the debugger, does not hang in interactive mode but hangs as
soon as STDOUT is redirected, you can have a hard time writing a bug
report.
> Digging up some history, I see that EU::MM does print the prompt and
> wait for an answer, but is that really the intended behavior or just a
> side-effect of the way scripted STDIN was implemented?
I cannot speak for MM but I'd call it sane behaviour.
> The 6.06_01
> Changes file simply says "prompt() will now read from STDIN.", but
> doesn't say whether it intended to redefine the defaulting behavior
> from "default" to "wait forever".
> perl -e 'use lib "6.05/lib";
> use ExtUtils::MakeMaker;
> warn ExtUtils::MakeMaker::prompt("foo", "bar"); ' > /tmp/prompt
> This simply took the default because STDOUT was not on the terminal. As
> of 6.06_01, it waits for input, but does that make it correct? (Or at
> least: is that they way all of the cool kids are using it these days?)
I'm just a bear of small brain but I would expect that he who asks a
question listens to my answer. And that he trusts that I am able to
handle the device that we are using to communicate.
> My take is that we have these four use-cases:
> o leave me alone: PERL_MM_USE_DEFAULT
> o ask me questions: -t STDIN and -t STDOUT
> o I can't hear you: -t STDIN and not -t STDOUT
> o harness/scripted: not -t STDIN and not -t STDOUT
Very nice summary. I'd label the third one with "capture" though.
> Does that cover it? If people *really* want this half-harness behavior,
> let's find SOMETHING that doesn't make "I can't hear you" behave so
> wrongly (at least put the prompt on both STDERR and STDOUT, though I
> would call that far from optimal.)
Writing to both STDOUT and STDERR is unacceptable. Asking a question
twice and expecting only one answer will hardly lead to a balance
between questions and answers.
> Regardless, the r9182 'fileno' change redefines the behavior WRT passing
> a missing default to prompt(). I think "we're out of answers" should
> be the same as when PERL_MM_USE_DEFAULT is set.
Don't know.
> $ svn up -r 9182 lib/Module/Build/Base.pm
> U lib/Module/Build/Base.pm
> Updated to revision 9182.
> $ echo "foo" | perl -e 'use Module::Build;
> warn Module::Build->prompt("foo", "bar");
> warn Module::Build->prompt("this");'
> foo [bar] foo at -e line 1, <STDIN> line 1.
> this
> Warning: something's wrong at -e line 1, <STDIN> line 1.
> $ PERL_MM_USE_DEFAULT=1 perl -e 'use Module::Build;
> warn Module::Build->prompt("foo", "bar");
> warn Module::Build->prompt("this");'
> foo [bar] bar
> bar at -e line 1.
> ERROR: This build seems to be unattended, but there is no default
> value for this question. Aborting.
> $ svn up -r 9181 lib/Module/Build/Base.pm
> U lib/Module/Build/Base.pm
> Updated to revision 9181.
> $ echo "foo" | perl -e 'use Module::Build;
> warn Module::Build->prompt("foo", "bar");
> warn Module::Build->prompt("this");'
> foo [bar] foo at -e line 1, <STDIN> line 1.
> ERROR: This build seems to be unattended, but there is no default
> value for this question. Aborting.
> $ svn up lib/Module/Build/Base.pm
> U lib/Module/Build/Base.pm
> Updated to revision 9211.
> $ echo "foo" | perl -e 'use Module::Build;
> warn Module::Build->prompt("foo", "bar");
> warn Module::Build->prompt("this");'
> foo [bar] foo at -e line 1, <STDIN> line 1.
> ERROR: This build seems to be unattended, but there is no default
> value for this question. Aborting.
> In any case, we need to print the prompt and *then* check STDIN or
> otherwise rearrange the manner in which we try to determine whether (@_
> == 1) is an error.
I'm pretty sure that you can get the behaviour you want without using
eof() at all, just with defined():
% echo "foo\nbar"|perl -le 'print defined <> ? "defined" : "undef" for 0..3'
defined
defined
undef
undef
--
andreas
