I was initially thinking this was a cost of memory vs. CPU scenario, but actually on 2nd thoughts, the CPU cycles is the same. So effectively we use less memory by default and any code that requires extra headers can parse them on a need-to-have basis.

Thanks for clearing up

Cheers

Jason

On Fri, Nov 4, 2011 at 10:49 AM, Daniel-Constantin Mierla <miconda@gmail.com> wrote:

Hi Jason,

I'm just continuing on a topic discussed few messages back in this thread.

Actually it is recommended to avoid adding new direct header hooks in sip_msg_t structure, because it becomes too big compared with what actually a SIP message has in terms of headers. Apart of invite which can have between 10-20 headers in average, the rest have less number of headers. Note that all headers are linked in a list held in sip_msg_t->headers.

Getting a header from the list is possible with several functions, depending whether you want the first or next header, searched by internal ID (if that header has one) or simply by name. The prototypes for these functions are in parser/msg_parse.h:

hdr_field_t* get_hdr(sip_msg_t *msg, enum _hdr_types_t ht);
hdr_field_t* next_sibling_hdr(hdr_field_t *hf);
hdr_field_t* get_hdr_by_name(sip_msg_t *msg, char *name, int name_len);
hdr_field_t* next_sibling_hdr_by_name(hdr_field_t *hf);

So getting my X-Header would be like:

parse_headers(msg, HDR_EOH_F, 0);
hdr = get_hdr_by_name(msg, "X-Header", strlen("X-Header"));

Some of them may be even removed from sip_msg_t (like priority or subject) which are not common at all.

Cheers,
Daniel

On 11/4/11 6:35 AM, Jason Penton wrote:
Hey Henning,

Ahh, thanks for that, thats perfect for now!

A pity, those names should have been a little more descriptive ;)

Thanks

Jason

On Thu, Nov 3, 2011 at 3:27 PM, Henning Westerholt <hw@kamailio.org> wrote:

On Thursday 03 November 2011, Jason Penton wrote:
> For Asserted and preferred identities, we don't need to parse the content,
> but in other headers I have not gotten to yet, we may need to.

Hi Jason,

do you talk about p-asserted and p-preferred header? This are fairly standard
headers, there are even some PVs to access them right now i think.

> Please help me understand, I would have thought from an architecture
> perspective, we would populate the sip_msg structure with all possible sip
> headers as well as the parsers. What is the reason we don't do this
> currently? performance?

I'd guess the reasons is memory efficiency. The structure get bigger and
bigger with every pointer. But for p-asserted and p-preffered, they are
already included it seems:

struct hdr_field* pai;
struct hdr_field* ppi;

Best regards,

Henning
_______________________________________________
sr-dev mailing list
sr-dev@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-dev
-- 
Daniel-Constantin Mierla -- http://www.asipto.com
Kamailio Advanced Training, Dec 5-8, Berlin: http://asipto.com/u/kat
http://linkedin.com/in/miconda -- http://twitter.com/miconda