[Ffmpeg-devel] cross compiling ffmpeg
Måns Rullgård
mru
Mon Feb 5 14:41:55 CET 2007
Marc Hoffman said:
>
> On Feb 5, 2007, at 8:28 AM, M?ns Rullg?rd wrote:
>
>>
>> Marc Hoffman said:
>>>
>>> On Feb 5, 2007, at 7:59 AM, Diego Biurrun wrote:
>>>
>>>> On Mon, Feb 05, 2007 at 06:49:19AM -0500, Marc Hoffman wrote:
>>>>>
>>>>> On Feb 5, 2007, at 3:57 AM, Diego Biurrun wrote:
>>>>>
>>>>>> I don't much like the echos and we don't add our names to
>>>>>> comments.
>>>>>> Also, you should try to hook up with the general structure of
>>>>>> configure and not hack in your own parts. There is one place where
>>>>>> we print output, there is one place where we handle options. Your
>>>>>> current patch is a quick hack that works for you but unacceptable
>>>>>> for
>>>>>> us unless you address the above issues.
>>>>>
>>>>> Thanks for helping me and sorry for the top posting.
>>>>>
>>>>> I'm not sure if I understand so help me through this ok. I made
>>>>> the
>>>>> change associated with my initials and echos. Which I agree with
>>>>> however. If you look at my change which is spread across the
>>>>> configure script. It was obvious to me that I need to:
>>>>>
>>>>> 1. modify the help
>>>>> 3. disable the default behavior of using uname to get targetos if
>>>>> targetos is not set
>>>>
>>>> OK
>>>>
>>>>> 2. insert the new argument handling
>>>>> 4. add default ignore handling to the main argument parsing loop.
>>>>
>>>> Not OK. After your patch there are two places where arguments are
>>>> handled. This is unacceptable.
>>>
>>> So would you prefer me moving all of the argument parsing to this new
>>> place this is fine by me but its a much larger change with lots of
>>> dependencies. And if I look at the configuration script it seems
>>> that there are 6 places 7 with my change that actually march over the
>>> command line in some way or another.
>>>
>>> 7 lines matching "for opt" in buffer configure.
>>> 237: for opt; do
>>> 243: for opt; do
>>> 249: for opt; do
>>> 255: for opt; do
>>
>> Those are doing something else. They only happen to use the same
>> variable name.
>>
>>> 632:for opt do
>>
>> This one you added.
>>
>>> 979:for opt do
>>
>> This could, and probably should, be merged with the main loop.
>>
>>> 991:for opt do
>>
>> The main loop.
>
> I have no problem should I move the entire argument parsing to this
> no location. I don't see any other way to make FFMPEG retargetable
> where the host os is different from the target os. I'm open to any
> suggestion.
Moving the loop as-is will break other things. Give me till tonight to think
about a good solution.
--
M?ns Rullg?rd
mru at inprovide.com
More information about the ffmpeg-devel
mailing list