On 1/07/2020 1:41 am, Barry Scott wrote:
On Saturday, 27 June 2020 07:47:14 BST Ian Haywood wrote:
I've been told in code review to use composition instead of inheritance,
which is fine in a general sense but I have difficulty applying to
twisted-specific task
thanks for these pointers eve
> On Jun 30, 2020, at 12:26 PM, meejah wrote:
>
> I recently wrote a relatively simple protocol to try this "sans-io" style
> and was happy to start with the state-machine (using Automat). This
> forced some good decisions. Then I worked outwards to the "more boring"
> bits (framing and serializ
As others have pointed out, you're stuck with a certain amount of
inheritance to use Twisted. However, this can be fairly minimal. For new
code trying to follow a "sans-io" approach is good too. The go-to
example of this for Twisted is Hyper/H2 and its integration into Twisted
Web for HTTP2 suppor
On Saturday, 27 June 2020 07:47:14 BST Ian Haywood wrote:
> In smb I have a SMBPacketReceiver that inherits from t.i.p.Protocol, it
> breaks the incoming TCP stream into logical packets (the analogue of
> LineReceiver in line-based protocols).
>
> I then subclass SMBPacketReceiver to SMBProtocol
On Sat, Jun 27, 2020 at 8:48 AM Ian Haywood wrote:
> In smb I have a SMBPacketReceiver that inherits from t.i.p.Protocol, it
> breaks the incoming TCP stream into logical packets (the analogue of
> LineReceiver in line-based protocols).
>
> I then subclass SMBPacketReceiver to SMBProtocol which d
In smb I have a SMBPacketReceiver that inherits from t.i.p.Protocol, it
breaks the incoming TCP stream into logical packets (the analogue of
LineReceiver in line-based protocols).
I then subclass SMBPacketReceiver to SMBProtocol which does a lot of the
"heavy lifting" analyzing incoming packet
