Re: error printing in reversed order ?

[nicolas bouillot <nicolas dot bouillot at gmail dot com>]

Thank you Joe and Dave.

I tried -fmax-errors but my test error (c++ iterator type) is itself
very long and still requires scrolling. In this case I had success
with tac. It just need to get some color back after filtering, which
is resulting for me in this following alias:
alias reversed_make='make 2>&1 >/dev/null | tac | egrep --color
"\b(error|cpp|hpp)\b|$ "'

Nicolas

On Fri, Oct 7, 2016 at 3:23 PM, David Malcolm <dmalcolm@redhat.com> wrote:
> On Fri, 2016-10-07 at 15:08 -0400, nicolas bouillot wrote:
>> Hi,
>>
>> Was wondering this could be a feature request ? Basically, this could
>> be a GCC option to print compilation errors in a reversed order, i.e.
>> the first being printed last. This is because when compiling from the
>> terminal, it would avoid mouse scrolling all day in order to get the
>> first error.
>>
>> I'll be happy to write a feature request somewhere if this deserves
>> it, but I do not know where and if this can be considered as a
>> feature
>> request.
>
> There's an option
>   -fmax-errors=
> which can be set to limit the number of errors emitted.
>
> It defaults to off.  The clang equivalent, -ferror-limit=, defaults to
> 20.
>
> Maybe we should change -fmax-errors= to default to on, maybe 20?  This
> ought to solve the "scrolling all day" problem you describe.
>
> Also, our error message is "computerese":
>
>   compilation terminated due to -fmax-errors=2.
>
> clang's is better:
>
>   fatal error: too many errors emitted, stopping now [-ferror-limit=]
>
> but we could improve ours by showing the number after the option), and
> say how to disable the limiter:
>
>   fatal error: too many errors emitted, stopping now [-fmax-errors=2]
>
> note: use -fmax-errors=0 to print all errors
>
> or somesuch.
>
> Thoughts?
>
> Dave