From: Nikolay Sivov <nsivov@codeweavers.com>
Subject: Re: [PATCH] xmllite: Avoid extra read from stream when identifying comment node.
Message-Id: <f87aa6c4-1fbe-a803-8ebc-fb6cf61979a9@codeweavers.com>
Date: Wed, 20 Nov 2019 13:32:21 +0300
In-Reply-To: <20191119014024.415408-1-whydoubt@gmail.com>
References: <20191119014024.415408-1-whydoubt@gmail.com>


On 11/19/19 4:40 AM, Jeff Smith wrote:
> Signed-off-by: Jeff Smith <whydoubt@gmail.com>
> ---
>   dlls/xmllite/reader.c       | 5 +++--
>   dlls/xmllite/tests/reader.c | 1 -
>   2 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/dlls/xmllite/reader.c b/dlls/xmllite/reader.c
> index eddc4d8eec..5299871136 100644
> --- a/dlls/xmllite/reader.c
> +++ b/dlls/xmllite/reader.c
> @@ -1459,7 +1459,7 @@ static HRESULT reader_parse_comment(xmlreader *reader)
>                       reader_init_strvalue(start, reader_get_cur(reader)-start, &value);
>                       TRACE("%s\n", debug_strval(reader, &value));
>   
> -                    /* skip rest of markup '->' */
> +                    /* skip rest of markup '-->' */
>                       reader_skipn(reader, 3);
That's obviously correct.
>   
>                       reader_set_strvalue(reader, StringValue_Value, &value);
> @@ -1472,8 +1472,9 @@ static HRESULT reader_parse_comment(xmlreader *reader)
>               }
>           }
>   
> -        reader_skipn(reader, 1);
>           ptr++;
> +        if (*ptr)
> +            reader_skipn(reader, 1);
>       }
I don't think it makes sense to change that just to get expected call 
sequence. Admittedly test itself is questionable.
>