[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: Problem finding telephonenumber in a plain numeric search when number is stored with special characters



Quanah Gibson-Mount wrote:



--On Sunday, August 28, 2005 10:35 PM +0200 Michael StrÃder <michael@stroeder.com> wrote:

Quanah Gibson-Mount wrote:


For Stanford, phone numbers are our second largest nightmare


;-)

people get to
enter their phone numbers into a web application, and you get very
interesting things like "N-ONE". :P


Well, if they really use one web application you can make your life
easier by immediately checking the input format there.


Unfortunately, it isn't that simple. Stanford is its own tel-co, which means we support all sorts of interesting things, like 5 digit numbers (first number is your stanford local prefix) and text numbers (5-HELP). So nearly anything is valid. The "none" was more accurately "5-NONE" or something like that, which is perfectly valid under our scheme, but not the persons real number. ;)

We do have a routine that strips out characters illegal under the telephoneNumber syntax, and rejects a few other things, but we are quite limited in what we can do.

I wouldn't find that weird for an APPROX matchingRule to turn aphas into digits for comparison. Something like a subfinal match would be a bit problematic, unless the matching rule can be fed with a notion of __default__ area code, so that shorter numbers can be compared to longer ones after appending the __default__ aera code; the default country code would also be required for an international directory. I think this could apply worldwide, except for those places where the leading "0" (or whatever) in the area code gets stripped when the country code is also present (Sweden for sure; don't know but many other countries likely).


p.


SysNet - via Dossi,8 27100 Pavia Tel: +390382573859 Fax: +390382476497