[MPlayer-dev-eng] [PATCH] trivial changes to suppress compiler warnings

[Reimar Döffinger]

On Thu, Oct 22, 2009 at 02:00:30PM +0200, Diego Biurrun wrote:
> Index: libmpcodecs/vf_pp7.c
> ===================================================================
> --- libmpcodecs/vf_pp7.c	(revision 29789)
> +++ libmpcodecs/vf_pp7.c	(working copy)
> @@ -286,8 +286,8 @@
>      int x, y;
>      const int stride= is_luma ? p->temp_stride : ((width+16+15)&(~15));
>      uint8_t  *p_src= p->src + 8*stride;
> -    DCTELEM *block= p->src;
> -    DCTELEM *temp= p->src + 32;
> +    DCTELEM *block= (DCTELEM*)p->src;
> +    DCTELEM *temp= (DCTELEM*)p->src + 32;

I think this breaks it.

> @@ -268,7 +268,7 @@
>  	struct rv_init_t init_data;
>  
>  	if(extrahdr_size < 8) {
> -	    mp_msg(MSGT_DECVIDEO,MSGL_ERR,"realvideo: extradata too small (%u)\n", sh->bih->biSize - sizeof(BITMAPINFOHEADER));
> +	    mp_msg(MSGT_DECVIDEO,MSGL_ERR,"realvideo: extradata too small (%lu)\n", sh->bih->biSize - sizeof(BITMAPINFOHEADER));

That's wrong e.g. on 64 Windows systems.
%zu probably would be right, but I find casting to int preferable.

> Index: stream/http.c
> ===================================================================
> --- stream/http.c	(revision 29789)
> +++ stream/http.c	(working copy)
> @@ -654,7 +654,7 @@
>  	mp_msg(MSGT_NETWORK,MSGL_V,"method:             [%s]\n", http_hdr->method );
>  	mp_msg(MSGT_NETWORK,MSGL_V,"status code:        [%d]\n", http_hdr->status_code );
>  	mp_msg(MSGT_NETWORK,MSGL_V,"reason phrase:      [%s]\n", http_hdr->reason_phrase );
> -	mp_msg(MSGT_NETWORK,MSGL_V,"body size:          [%d]\n", http_hdr->body_size );
> +	mp_msg(MSGT_NETWORK,MSGL_V,"body size:          [%zu]\n", http_hdr->body_size );

IMO body_size and buffer_size shouldn't be size_t in the first place, though
that would need to check in detail if it is correct or would cause more issues.

> Index: libvo/x11_common.c
> ===================================================================
> --- libvo/x11_common.c	(revision 29789)
> +++ libvo/x11_common.c	(working copy)
> @@ -1341,7 +1341,7 @@
>  
>  void vo_x11_fullscreen(void)
>  {
> -    int x, y, w, h;
> +    int x=0, y=0, w=0, h=0;

The whole code of this function should probably be written in a better
way, but here specifically I'd suggest just moving the
            x = vo_old_x;
            y = vo_old_y;
            w = vo_old_width;
            h = vo_old_height;
part up here.

> Index: loader/qtx/qtxsdk/components.h
> ===================================================================
> --- loader/qtx/qtxsdk/components.h	(revision 29789)
> +++ loader/qtx/qtxsdk/components.h	(working copy)
> @@ -728,15 +728,15 @@
>      if(cd->wantedDestinationPixelTypes){
>  	unsigned int* p=cd->wantedDestinationPixelTypes;

Doesn't that assignment cause a waning, too?

>      printf("screenFloodMethod=%d value=%d  preferredOffscreenPixelSize=%d\n",
>  	cd->screenFloodMethod, cd->screenFloodValue, cd->preferredOffscreenPixelSize);
>      printf("callbacks: progress=%p compl=%p data=%p ftime=%p srcdata=%p sync=%p\n",
> -     cd->progressProcRecord, cd->completionProcRecord,
> -     cd->dataProcRecord, cd->frameTime, cd->sourceData, cd->syncFrameTime);
> +     (void*)cd->progressProcRecord, (void*)cd->completionProcRecord,
> +     (void*)cd->dataProcRecord, cd->frameTime, cd->sourceData, cd->syncFrameTime);

I think the format string should be changed instead.

> Index: libmpdemux/demux_lmlm4.c
> ===================================================================
> --- libmpdemux/demux_lmlm4.c	(revision 29789)
> +++ libmpdemux/demux_lmlm4.c	(working copy)
> @@ -72,7 +72,7 @@
>      if ( frame->channelNo > 7 ||
>           frame->frameSize > MAX_PACKET_SIZE || frame->frameSize <= 0)
>      {
> -        mp_msg(MSGT_DEMUX, MSGL_V, "Invalid packet in LMLM4 stream: ch=%d size=%d\n", frame->channelNo, frame->frameSize);
> +        mp_msg(MSGT_DEMUX, MSGL_V, "Invalid packet in LMLM4 stream: ch=%d size=%zd\n", frame->channelNo, frame->frameSize);
>          return 0;
>      }
>      switch (frame->frameType) {
> @@ -156,7 +156,7 @@
>      frameInfo->frameSize = packetSize - 8; //sizeof(IME6400Header);
>      frameInfo->paddingSize = (packetSize & PACKET_BLOCK_LAST) ? PACKET_BLOCK_SIZE - (packetSize & PACKET_BLOCK_LAST) : 0;
>  
> -    mp_msg(MSGT_DEMUX, MSGL_DBG2, "typ: %d chan: %d size: %d pad: %d\n",
> +    mp_msg(MSGT_DEMUX, MSGL_DBG2, "typ: %d chan: %d size: %zd pad: %zd\n",
>              frameInfo->frameType,
>              frameInfo->channelNo,
>              frameInfo->frameSize,

Using size_t for frameSize IMO is a bad idea.
Same for the rest of those changes.