I took a quick look at your header file, and a couple of things spring out:
* There seem to be a lot of functions there. I suspect that many of them
are used only internally to the library, and hence shouldn't be in the
exported header file.
* The fuctions do not all live in the same namespace -- some start with
UTIL, some with SPF, MTA, ERROR, MACRO. It is much more preferable to
have all the externally visible functions start with the same prefix.
* the SPF_PREFIX enum has a bogus ,T in it
* It is typically bad practice to have an initialized array in a header
file (unless the use is very carefully controlled).
* in your peer_info_t, it is not clear that all the fields have the
correct length. As I recall, domain names can be up to 256 characters.
Philip
James Couzens wrote:
> http://libspf.org/files/spf.h
>
> Here is the header from our alpha release.
>
> If you are running Qmail, and you are willing to test the alpha version
> drop me a line and I'll see you get a copy. Currently there are memory
> problems in the macro parsing functions so parsing records of those
> nature currently are handled by a stub function that simply allocates
> memory equal to the passed "macro" and returns it to the calling
> function.
>
> Cheers,
>
> James
>
> On Tue, 2004-01-20 at 08:16, Terence Way wrote:
>
>>If you guys are feeling, er, critical:
>>
>>Attached is SPF.H, a proposed API into an asynchronous, nonblocking SPF
>>implementation.
>>If the attachment doesn't go through, download it at
>>http://www.wayforward.net/spf/spf.h
>>
>>The only implementation I have behind this is throw-away code that
>>checks if this API works with both libresolv and adns.
>>
>>Summary:
>>spf_init()
>> called once at program startup (might call res_init, for instance)
>>
>>spf_open()
>> create a context for an asynchronous query, context contains DNS
>>queries for the user to perform.
>>
>>spf_txt()
>>spf_a()
>>spf_ptr()
>>spf_mx()
>> called for each answer/additional-info record in a DNS response.
>>Example: two MX records with corresponding A records in the
>>additional-info would result in spf_mx(), spf_mx(), spf_a(), spf_a().
>>
>>spf_check()
>> Steps through the SPF checker, returns spf_incomplete if there are
>>more DNS queries for the user to perform, or returns one of spf_fail,
>>spf_pass, spf_unknown, spf_error.
>>
>>spf_close()
>> Free up the SPF context.
>>
>>There is a high-level function:
>>spf_query()
>> which uses a built-in DNS library and blocks until the SPF query is
>>complete.
>>
>>Design notes:
>>spf_open() will merely allocate necessary storage.
>>
>>The spf_query structure returned by spf_open() and used by the other
>>functions keeps a miniature DNS cache of all DNS records fetched by the
>>user. The functions spf_txt(), spf_a(), spf_ptr(), and spf_mx() merely
>>add information to that cache.
>>
>>The real work is done by spf_check(). Its pseudo-code:
>> if already done:
>> return previous result
>> if no SPF text:
>> look in cache for TXT record
>> if not found
>> return spf_unknown
>> find 'v=spf1'
>> if none, or more than one,
>> return spf_unknown
>> tokenize text
>> set current token pointer to first token
>> while we have more tokens:
>> examine current token pointer
>> if we have information in DNS cache:
>> do check
>> if match, return match status
>> otherwise advance current token pointer
>> else
>> set up for DNS query
>> return spf_incomplete
>>
>>The cache records all DNS results, and pays no attention to TTL.
>>
>>More later.
>>T. Way
>>terry@wayforward.net
>>
>>--
>>To unsubscribe, change your address, or temporarily deactivate your subscription,
>>please go to http://v2.listbox.com/member/?listname@Ë`Ì{5¤¨wâÇSÓ°)h
-------
To unsubscribe, change your address, or temporarily deactivate your subscription,
please go to http://v2.listbox.com/member/?listname@Ë`Ì{5¤¨wâÇSÓ°)h
* There seem to be a lot of functions there. I suspect that many of them
are used only internally to the library, and hence shouldn't be in the
exported header file.
* The fuctions do not all live in the same namespace -- some start with
UTIL, some with SPF, MTA, ERROR, MACRO. It is much more preferable to
have all the externally visible functions start with the same prefix.
* the SPF_PREFIX enum has a bogus ,T in it
* It is typically bad practice to have an initialized array in a header
file (unless the use is very carefully controlled).
* in your peer_info_t, it is not clear that all the fields have the
correct length. As I recall, domain names can be up to 256 characters.
Philip
James Couzens wrote:
> http://libspf.org/files/spf.h
>
> Here is the header from our alpha release.
>
> If you are running Qmail, and you are willing to test the alpha version
> drop me a line and I'll see you get a copy. Currently there are memory
> problems in the macro parsing functions so parsing records of those
> nature currently are handled by a stub function that simply allocates
> memory equal to the passed "macro" and returns it to the calling
> function.
>
> Cheers,
>
> James
>
> On Tue, 2004-01-20 at 08:16, Terence Way wrote:
>
>>If you guys are feeling, er, critical:
>>
>>Attached is SPF.H, a proposed API into an asynchronous, nonblocking SPF
>>implementation.
>>If the attachment doesn't go through, download it at
>>http://www.wayforward.net/spf/spf.h
>>
>>The only implementation I have behind this is throw-away code that
>>checks if this API works with both libresolv and adns.
>>
>>Summary:
>>spf_init()
>> called once at program startup (might call res_init, for instance)
>>
>>spf_open()
>> create a context for an asynchronous query, context contains DNS
>>queries for the user to perform.
>>
>>spf_txt()
>>spf_a()
>>spf_ptr()
>>spf_mx()
>> called for each answer/additional-info record in a DNS response.
>>Example: two MX records with corresponding A records in the
>>additional-info would result in spf_mx(), spf_mx(), spf_a(), spf_a().
>>
>>spf_check()
>> Steps through the SPF checker, returns spf_incomplete if there are
>>more DNS queries for the user to perform, or returns one of spf_fail,
>>spf_pass, spf_unknown, spf_error.
>>
>>spf_close()
>> Free up the SPF context.
>>
>>There is a high-level function:
>>spf_query()
>> which uses a built-in DNS library and blocks until the SPF query is
>>complete.
>>
>>Design notes:
>>spf_open() will merely allocate necessary storage.
>>
>>The spf_query structure returned by spf_open() and used by the other
>>functions keeps a miniature DNS cache of all DNS records fetched by the
>>user. The functions spf_txt(), spf_a(), spf_ptr(), and spf_mx() merely
>>add information to that cache.
>>
>>The real work is done by spf_check(). Its pseudo-code:
>> if already done:
>> return previous result
>> if no SPF text:
>> look in cache for TXT record
>> if not found
>> return spf_unknown
>> find 'v=spf1'
>> if none, or more than one,
>> return spf_unknown
>> tokenize text
>> set current token pointer to first token
>> while we have more tokens:
>> examine current token pointer
>> if we have information in DNS cache:
>> do check
>> if match, return match status
>> otherwise advance current token pointer
>> else
>> set up for DNS query
>> return spf_incomplete
>>
>>The cache records all DNS results, and pays no attention to TTL.
>>
>>More later.
>>T. Way
>>terry@wayforward.net
>>
>>--
>>To unsubscribe, change your address, or temporarily deactivate your subscription,
>>please go to http://v2.listbox.com/member/?listname@Ë`Ì{5¤¨wâÇSÓ°)h
-------
To unsubscribe, change your address, or temporarily deactivate your subscription,
please go to http://v2.listbox.com/member/?listname@Ë`Ì{5¤¨wâÇSÓ°)h
