[x264-devel] x264-devel Digest, Vol 114, Issue 5

[Henrik Gramner]

On Mon, Nov 7, 2016 at 7:58 PM, Wim Rijnders <wrijnders at gmail.com> wrote:
> I tested with the latest git of x264 and indeed cli_input is not a required
> symbol any more.
> So this has been fixed in the later versions of x264. I see reason to
> upgrade.
>
> But why are you puzzled? I can see the fix in the makefile. A better
> response is 'this has been fixed'.

I wasn't aware of any recently fixed linking issues related to
cli_input, had I known about it I would have pointed it out of course.

I'm not able to find this supposed fix however. The error message you
posted earlier indicates you were using a revision with API version
148 and you're saying there's a fix in the Makefile.

API version 148 was introduced 2015-07-25 and there has been 9 changes
to the Makefile since then
https://git.videolan.org/?p=x264.git;a=history;f=Makefile but none of
those commits address any linking issues.

>> Why the compiler
>> decides to make that line in particular trigger a warning/error is a
>> more interesting question.
>
> From my first message:
>
> "I understand that this is due to the very strict checks I enabled on my
> project."

Yes, my point is I'd like to know why the compiler only warns about
_this_ very specific line and not any of the hundreds(?) of other
implicit casts in the code base. It could indicate a genuine issue or
it's just a quirk in the compiler - either way we need to understand
the underlying reason before considering doing anything about it. Just
randomly adding casts to silence warnings isn't considered good
development practice.

> It worries me that you have no appreciation of this. I sense that I have no
> reason to be here.
>
> goodbye and thank you for your time.

It's unfortunate that you feel that way, but thank you for the report
and best of luck in the future.