On Tue, 28 Jul 2009, Francesco Saverio Giudice wrote:
Hi,
> As I can see in original inet.c, and if I don't read wrongly, iLen (that is
> @nResult var) it is the length of received string or 0 or error, but in
> case of length of the string it will contains also "\r\n" length. Instead
> iPos is used to search and return the length of the string without pattern.
> So it seems to me that on this point docs are correct also if not reports
> -2 as possible error (timeout).
No it isn't.
The value stored in 2-nd parameter passed by reference is length of string
_WITHOUT_ CRLF (or other line terminator in hb_inetRecvEndBlock()) when
documentation says it should be 2 or more and 0 should indicate end of
connection (connection closed by peer).
Now in Harbour and xHarbour in old and current code in second parameter
is stored size of string without line (string) terminator so it's exactly
the len() of return line.
Just make a test:
// simple server code
proc main()
local nPort, sock, con, n
nPort := 2345
n := 0
sock := hb_inetServer( nPort )
while .t.
con := hb_inetAccept( sock )
hb_inetSend( con, "This is line: " + hb_ntos( ++n ) + chr(13)+chr(10) )
hb_inetSend( con, "EOF" + chr(13)+chr(10) )
hb_inetClose( con )
con := NIL
enddo
return
// simple client code
proc main()
local cAddr, nPort, sock, cLine, nResult
cAddr := "127.0.0.1"
nPort := 2345
sock := hb_inetConnectIP( cAddr, nPort )
cLine := hb_inetRecvLine( sock, @nResult )
? "hb_inetRecvLine()"
? "line:", cLine
? "result:", nResult, len( cLine )
?
cLine := hb_inetRecvEndBlock( sock, { chr(13)+chr(10) }, @nResult )
? "hb_inetRecvEndBlock()"
? "line:", cLine
? "result:", nResult, iif( valtype( cLine ) == "C", len( cLine ), "" )
?
return
> I suppose that Giancarlo uses NIL as return string in case of error to use
> function without checking nResult. An empty string forces to check always
> nResult.
I'm afraid he described sth what he plan to make but the real code never
reached the documentation conditions or maybe it was updated later and
documentation not but for sure it does not work as expected. In xHarbour
it's even still buggy in internal conditions.
Now we have question what we should fix. Documentation or code.
Please also remember that we should keep hb_inetRecvEndBlock() synced.
We should report at least 3-states.
1. received new line
2. connection closed by forign host
3. error
Original documentation addresses above three states so maybe we should
implement it. In such case the return value is less important and we can
return empty string on errors or even NIL if someone prefers it. Anyhow
please note that now this functions can return on error 0, NIL or ""
and I would like to eliminate it and precisely define returned values.
best regards,
Przemek
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
