Don't bait me Mr. Swift! :-)
Gary
On Sat, Jul 22, 2017 at 12:54 PM, Matt Sicker wrote:
> I usually stick to big endian encoding while on the JVM thanks to
> ByteBuffer making it easy. If you write your own byte buffer style class,
> then it doesn't really matter which endianness you choose as y
I usually stick to big endian encoding while on the JVM thanks to
ByteBuffer making it easy. If you write your own byte buffer style class,
then it doesn't really matter which endianness you choose as you'll have to
split up larger primitives into bytes regardless.
On 21 July 2017 at 15:10, Gary G
The trick is to define "length": what is the size and endianness for this
integer and are either configurable? I think JMeter has some code for that.
But at that point, why not use the BSON for uint64 (little-endian ?
Gary
On Fri, Jul 21, 2017 at 12:43 PM, Matt Sicker wrote:
> Exactly! I've se
Exactly! I've seen similar encodings used in simple TCP framing. It should
be robust enough for logging (no need for additional metadata).
On 21 July 2017 at 14:39, Gary Gregory wrote:
> Like Pascal strings, lengh-data-length-data...
>
> On Fri, Jul 21, 2017 at 11:31 AM, Matt Sicker wrote:
>
>
Like Pascal strings, lengh-data-length-data...
On Fri, Jul 21, 2017 at 11:31 AM, Matt Sicker wrote:
> If the size of the message is known ahead of time, you can create a simple
> framing protocol around messages as well.
>
> On 21 July 2017 at 07:23, Mikael Ståldal wrote:
>
> > I don't quite un
If the size of the message is known ahead of time, you can create a simple
framing protocol around messages as well.
On 21 July 2017 at 07:23, Mikael Ståldal wrote:
> I don't quite understand.
>
> This is something which is needed now. When you use a streaming appender
> (Console, any File or So
I don't quite understand.
This is something which is needed now. When you use a streaming appender
(Console, any File or Socket with protocol=tcp/ssl), the layout needs to
provide some way to delimit log events.
On 2017-07-20 22:28, Matt Sicker wrote:
I'm not a fan of the null termination a
I'm not a fan of the null termination as it is. Do you know how many
millions or billions of dollars have been wasted due to buggy C and C++
programs thanks to null-terminated strings? Almost as bad as null itself. ;)
Having some way to delimit log events could be handy later on for
garbage-free n
Now JsonLayout, XmlLayout, YamlLayout and GelfLayout has an
includeNullDelimiter option to terminate each log event with a 0-byte.
Would it be useful to have the same in any other layout?
I could think of HtmlLayout, SyslogLayout and RFC5424Layout.
