pre-RFR (s): 8049847: Enhance PrintWriter line.separator handling

Xueming Shen xueming.shen at oracle.com
Sun Jan 4 21:37:22 UTC 2015


On 1/4/15 12:25 PM, Alan Bateman wrote:
> On 04/01/2015 19:18, Xueming Shen wrote:
>> :
>>
>> I agree with Alan that the general interface "Appendable" probably 
>> should not get
>> into the specific "formatting" functionality here. And I also think 
>> the "line.separator"
>> should not be in the "Writer" as well, writer/reader does not deal 
>> with "formatting",
>> only couple "special" writers need to work with "line" 
> I've often wondered about BufferedReader/BufferedWriter supporting 
> lines, it often feels like the readLine and newLine methods should 
> have been further up in the hierarchy in Reader and Writer so the 
> BufferedXXX types just adds buffering. I realize of course that 
> newLine might be very inefficient without buffering. Do have all the 
> history? I'm wondering if there is merit looking at this again as it 
> could be useful beyond the case that Claes wants to address. If Writer 
> did define newLine then then the proposed lineSeperator wouldn't be 
> needed.
>
> -Alan

I don't have the history. My take on this is that the read/newLine() 
methods in BufferedReader/Writer
are more a pair of utility/convenient/bonus method. They serve parsing 
and formatting functionality,
which probably should belong to the Scanner and Formatter, rather than 
the io reader and writer. Of
course we didn't have scanner and formatter in earlier days:-) And it's 
convenient to have them in
the buffered reader/writer, if the line is only thing you care about on 
top of "character". When I care
about a little more than a "line", I started to use Scanner more often 
these days.

-Sherman



More information about the core-libs-dev mailing list