[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH 0 of 6 V1] Counters
[Thread Prev] | [Thread Next]
- Subject: Re: [PATCH 0 of 6 V1] Counters
- From: Aris Adamantiadis <aris@xxxxxxxxxxxx>
- Reply-to: libssh@xxxxxxxxxx
- Date: Tue, 11 Feb 2014 10:55:43 +0100
- To: libssh@xxxxxxxxxx
- Cc: audrius.butkevicius@xxxxxxxxx
Hi Audreus, Le 10/02/14 17:35, Audrius Butkevicius a écrit : > Hi Aris, > > On 10/02/2014 16:00, Aris Adamantiadis wrote: >> - Doxygen comments: >> You comments are great. Could you put them in the implementation of >> functions (counters.c) as we do for the other includes in libssh.h ? >> That would be a little more coherent. > I assume you mean put comments in counters.c as well as in libssh.h? I mean moving all doxygen comments for functions in counters.c. If there are doxygen comments for the structures, they can stay in libssh.h. >> On the code itself: >> - I think you can merge ssh_bytes_counter_struct and >> ssh_packet_counter_struct. It's very likely that both will be used at >> same time and that reduces the overall complexity. ssh_counter_struct is >> also shorter :) Also means less setters. >> What do you think ? > I can do this, but the issue with this is that in some cases there will > be no bytes, or no packets. That's not a big problem. we can always find data sources for these counters. > Furthermore, we can only reduce (for the session) by one argument, > because you still need to have separate counters for bytes/packets on > the socket, as well as bytes (but not packets) prior the protocol overhead. > Meaning that I'd still have to store two ssh_counter structs in the > session, and for the channel I'd be holding a struct which only uses > bytes (but not packets). Agreed but two pointers is better than 4. Let's keep it simple. Aris > > > Thanks, > Audrius. >
Re: [PATCH 0 of 6 V1] Counters | Aris Adamantiadis <aris@xxxxxxxxxxxx> |
Re: [PATCH 0 of 6 V1] Counters | Audrius Butkevicius <audrius.butkevicius@xxxxxxxxxxxxxxxx> |