Re: [PATCH] perf tool: reset use_browser when compiled without NEWT

[David Ahern <dsahern@gmail.com>]

On 11/28/2011 10:09 AM, Arnaldo Carvalho de Melo wrote:
> Em Mon, Nov 28, 2011 at 09:58:46AM -0700, David Ahern escreveu:
>> Hi Arnaldo:
>>
>> didn't see this one in your queue or a response. Do you agree this is
>> needed?
> 
> Well, I think we should provide a better explanation for this usecase,
> i.e. the code as is will use --stdio in NO_NEWT=yes builds, but will
> still offer --tui and if that is specified, will produce no output.
> 
> Perhaps it is best to either elide --tui in NO_NEWT builds or state that
> the build doesn't support --tui more clearly.
> 
> Or are you seeing something else that you think needs fixing?
>  
> - Arnaldo

In my case I compiled the code on 32-bit RHEL5 with newt disabled. 'perf
report' showed no output unless I added --stdio which seemed silly since
newt was disabled. Adding the  'use_browser = 0' to the setup_browser
fixed it. Doing that mirrors what is done in util/ui/setup.c,
setup_browser() which is used when newt is not disabled compile time:

    if (!isatty(1) || !use_browser || dump_trace) {
        use_browser = 0;
        if (fallback_to_pager)
            setup_pager();
        return;
    }