Hi,

2011/8/8 Måns Rullgård <[email protected]>:
> "Ronald S. Bultje" <[email protected]> writes:
>> On Mon, Aug 8, 2011 at 11:17 AM, Diego Biurrun <[email protected]> wrote:
>>> On Mon, Aug 08, 2011 at 11:00:40AM -0700, Ronald S. Bultje wrote:
>>>> On Mon, Aug 8, 2011 at 3:30 AM, Diego Biurrun <[email protected]> wrote:
>>>> > --- a/libavcodec/truespeech.c
>>>> > +++ b/libavcodec/truespeech.c
>>>> > @@ -33,13 +33,13 @@
>>>> >  */
>>>> >  typedef struct {
>>>> >     /* input data */
>>>> > -    int16_t vector[8];  //< input vector: 5/5/4/4/4/3/3/3
>>>> > -    int offset1[2];     //< 8-bit value, used in one copying offset
>>>> > -    int offset2[4];     //< 7-bit value, encodes offsets for copying 
>>>> > and for two-point filter
>>>> > -    int pulseoff[4];    //< 4-bit offset of pulse values block
>>>> > -    int pulsepos[4];    //< 27-bit variable, encodes 7 pulse positions
>>>> > -    int pulseval[4];    //< 7x2-bit pulse values
>>>> > -    int flag;           //< 1-bit flag, shows how to choose filters
>>>> > +    int16_t vector[8];  // input vector: 5/5/4/4/4/3/3/3
>>>> > +    int offset1[2];     // 8-bit value, used in one copying offset
>>>> > +    int offset2[4];     // 7-bit value, encodes offsets for copying and 
>>>> > for two-point filter
>>>> > +    int pulseoff[4];    // 4-bit offset of pulse values block
>>>> > +    int pulsepos[4];    // 27-bit variable, encodes 7 pulse positions
>>>> > +    int pulseval[4];    // 7x2-bit pulse values
>>>> > +    int flag;           // 1-bit flag, shows how to choose filters
>>>> >     /* temporary data */
>>>> >     int filtbuf[146];   // some big vector used for storing filters
>>>> >     int prevfilt[8];    // filter from previous frame
>>>> > --- a/libavcodec/vp3.c
>>>> > +++ b/libavcodec/vp3.c
>>>> > @@ -225,7 +225,7 @@ typedef struct Vp3DecodeContext {
>>>> >     /* these arrays need to be on 16-byte boundaries since SSE2 
>>>> > operations
>>>> >      * index into them */
>>>> > -    DECLARE_ALIGNED(16, int16_t, qmat)[3][2][3][64];     
>>>> > //<qmat[qpi][is_inter][plane]
>>>> > +    DECLARE_ALIGNED(16, int16_t, qmat)[3][2][3][64];     // 
>>>> > qmat[qpi][is_inter][plane]
>>>> > --- a/libavfilter/vf_drawbox.c
>>>> > +++ b/libavfilter/vf_drawbox.c
>>>> > @@ -34,7 +34,7 @@ enum { Y, U, V, A };
>>>> >  typedef struct {
>>>> >     int x, y, w, h;
>>>> >     unsigned char yuv_color[4];
>>>> > -    int vsub, hsub;   //< chroma subsampling
>>>> > +    int vsub, hsub;   // chroma subsampling
>>>> >  } DrawBoxContext;
>>>>
>>>> All these are struct members, are you sure you want to drop their doxy?
>>>
>>> I see little point in (doxygen) documenting structs only used in a
>>> single file and/or documenting only single struct members.  But I
>>> can surely be convinced of something else...
>>
>> We do that in many decoders. It helps to document the implementation
>> if ever, in the far far future ahead, the original author disappears
>> and some poor new soul has to fix bugs or add new features instead.
>
> Plain comments are good enough for that.  Doxygen is mostly intended to
> generate documentation for people who otherwise have no reason to read
> the code.

The idea is "have documentation open in browser, and look at actual
code in editor". Otherwise you need multiple code editors open.

Ronald
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to