[PATCH 5/5] Limit the scope of gasidx in add_plan_to_notes

[Lubomir I. Ivanov]

On 11 July 2014 16:46, Tomaz Canabrava <tcanabrava at kde.org> wrote:
>
> Em 11/07/2014 10:05, "Lubomir I. Ivanov" <neolit123 at gmail.com> escreveu:
>
>
>>
>> On 11 July 2014 14:55, Anton Lundin <glance at acc.umu.se> wrote:
>> > On 11 July, 2014 - Lubomir I. Ivanov wrote:
>> >
>> >> On 11 July 2014 10:13, Anton Lundin <glance at acc.umu.se> wrote:
>> >> > glance> for (int i=0;i<3;i++) ?
>> >>
>> >> for that to work on older versions e.g. MSVC 2003, /Tp has to be used
>> >> (like jeff mentioned, we can mark the input to be C++) and then extern
>> >> "C" wrapping would be needed to preserve the binary interface.
>> >>
>> >> on even older versions like MSVC 1998, /Tp exists but the actual
>> >> intermix feature from c99 is a bit messed up. the lexical scanner does
>> >> recognize it, but then it assigns a non-local scope to the iterator.
>> >> so you can do things like:
>> >>
>> >> for (int i = 0; i < 10; i++)
>> >>     puts("stuff");
>> >> i = 0; // <---
>> >>
>> >> i would advice against this particular feature if targeting older
>> >> MSVC, but for third party source projects i usually follow their usage
>> >> of "for" loops.
>> >> for my own managed projects, this:
>> >>
>> >> int i = 0;
>> >> while (i < ...) {... i++ }
>> >>
>> >> remained as a habit.
>> >>
>> >
>> > Msvc 2013 seems to get it right, and i can gladly argue that we
>> > shouldn't support any older Msvc's just for the sake of it.
>> >
>> > How many do actually build subsurface for windows? You, Dirk and maybe 1
>> > 2 other dev's, but thats probably it. Dirk uses mingw, and you would
>> > know to use msvc 2013, so it doesn't feel like a show stopper for me =)
>> >
>> > But, if we as always, if we know that things break with older msvc's we
>> > should document it.
>> >
>>
>> well, my opinion is that mingw is sufficient of a complier for windows
>> and in the case of subsurface.
>> MSVC support would be a time consuming exercise with no real benefits.
>
> Msvc generates faster executables on win32 than gcc, this is a benefit.
>

i was going to mention that as it used to be my own experience from
the early 2000+ years, but then looked at this:
http://stackoverflow.com/questions/14734068/mingw-4-7-2-vs-visual-c-11-0-2012-benchmark

so unless we measure it, it's arguable.

lubomir
--