- From: Daniel Brockman <daniel@brockman.nu>
- Date: Tue, 30 Sep 2003 02:03:32 +0200
- To: "Ted Wugofski" <ted.wugofski@openwave.com>
- Cc: <Bennett.Marks@nokia.com>, <uri@w3.org>
> > I don't see how mobile devices can't do the same---i.e., having "Send > > MMS" as an action for, e.g., hyperlinks or selected text. > > The purpose of the "mmsto:" URI is so that a browser (or some other > client) knows to process the link using the MMS composer application. > > [...] Of course. I was trying to critique the way Bennett apparently planned to use it ("making a resource forwardable via MMS," as opposed to "making someone contactable via MMS"---compare, for the mailto scheme, "making some text sendable via e-mail" vs. "making someone contactable via e-mail"). However, on a second thought, nevermind. It's not very important, as I'm sure everybody will just use their best judgement and everything will turn out alright in the end. > > I agree. The "mms" vs "mmsto" distinction seems confusing at best. > > Again, it may just be that I don't see the right use cases for it. I > > think the purpose needs to be clarified. > > See my response to Bennett. Hopefully that clears it up. The "mms" URI > scheme is used for applications to send MMS message to other > applications -- likely without direct user interaction. In essence, the > "mms" URI scheme should behave like the "http" URI scheme ala this quote > from RFC 2368: > > In current practice, resolving URLs such as those in the "http" > scheme causes an immediate interaction between client software and a > host running an interactive server. > > This is the behaviour I would expect when processing an "mms" URI. But > again, since it allows for you to attache content, I can't help but > think it is referring to the message more than the actual recipient. By this logic, there should be two separate http schemes: one for end users and one for web services. Right? Why can't the mobile device in your game example simply provide two API functions: "SendMms" and "OpenMmsComposer"? I think this would be much less confusing than I guess having a "InterpretUri" function that would either send a message directly or open up a composer, depending on the URI itself. -- Daniel Brockman daniel@brockman.nu ----- Original Message ----- From: "Ted Wugofski" <ted.wugofski@openwave.com> To: "Daniel Brockman" <daniel@brockman.nu> Cc: <Bennett.Marks@nokia.com>; <uri@w3.org> Sent: Monday, September 29, 2003 11:29 PM Subject: Re: IETF Draft on mmsto: > > > I don't see how mobile devices can't do the same---i.e., having "Send > > MMS" as an action for, e.g., hyperlinks or selected text. > > The purpose of the "mmsto:" URI is so that a browser (or some other > client) knows to process the link using the MMS composer application. > > > I merely want to point out that using the mmsto scheme to identify > > an *action* seems abusive, since the scheme obviously is meant to > > identify MMS destinations, > > Actually, nope. The purpose of the "mmsto" URI scheme is to identify a > _message_ that may be sent to an MMS recipient -- thats why the "mmsto:" > URI scheme allows for optional header values. > > In the I-D, I talk about identifying the device via several mechanisms > RFC 3601 & RFC 2822 being two of them. > > (Yes, I know what 2368 says, but I think that with _current practice_, > "mailto:" has come to mean the message going over the wire rather than > the mailbox since any browser or email client I use these days allows me > to compose prior to sending or to cancel altogether) > > > Granted, the mailto scheme with its subject and body parameters > > currently identifies some kind of actions, but I don't agree with this > > either. > > Looks like we will have to disagree about the role of "mmsto" as I am > modeling it after "mailto". > > > I agree. The "mms" vs "mmsto" distinction seems confusing at best. > > Again, it may just be that I don't see the right use cases for it. I > > think the purpose needs to be clarified. > > See my response to Bennett. Hopefully that clears it up. The "mms" URI > scheme is used for applications to send MMS message to other > applications -- likely without direct user interaction. In essence, the > "mms" URI scheme should behave like the "http" URI scheme ala this quote > from RFC 2368: > > In current practice, resolving URLs such as those in the "http" > scheme causes an immediate interaction between client software and a > host running an interactive server. > > This is the behaviour I would expect when processing an "mms" URI. But > again, since it allows for you to attache content, I can't help but > think it is referring to the message more than the actual recipient. > > > > > Daniel Brockman wrote: > > > Hi Bennet, > > > > > >>For the email scheme, far and away the most common usage pattern is the > > > > reply, and hence the email address is in the primary (mandatory) position > > before the "?". This is one common pattern for mobile, but the other equally > > common pattern for sms and mms is forwarding a piece of fixed information > > (i.e. a URL or image) to a new party. In this case the universal mode for > > this is to have the composer "find" the address in the address book. The > > current specification for mmsto: would require a default telephone # or > > email address (the reply scenario), and make it non-intuitive for the > > forwarding case, where I would pre-specify the "body" but not the address. > > It would require the user to explicitly overwrite a bogus address supplied > > with the scheme, and if not overwritten, the message would go back to the > > default location, definitely not the desired result. > > > > What you're talking about is "the action of sending x", where x is some > > resource---presumably identified by a URI. I don't know if this action > > should be identifed by a URI, but if it should, I don't think the scheme > > should be mmsto. Rather, we should define a new scheme, let's call it > > "send," that would identify the action of sending a resource to someone by > > some means. So for instance you could have > > > > send:http://www.example.com/?to=mailto:someome@example.com > > > > or---to take your example of sending it to some unspecified > > destination---simply > > > > send:http://www.example.com/ > > > > or you could even send it to an FTP server: > > > > send:http://www.example.com/?to=ftp://ftp.example.com/upload/ > > > > But we're lived without the send scheme for---what?---almost 20 years now. > > How come we managed? Well, I don't think this action should be explicitly > > identified by an URI, just like I don't think clicking a button should not > > be identified by an URI. Simply add a "Send" option to the browser, > > applicable to resources which can be identified by, e.g., a hyperlink, a > > block of selected text, or the current document. In fact, this is exactly > > what browsers do today, for e-mail. I don't see how mobile devices can't do > > the same---i.e., having "Send MMS" as an action for, e.g., hyperlinks or > > selected text. > > > > To clarify: I'm not suggesting that "send" actually be defined as a new URI > > scheme; I merely want to point out that using the mmsto scheme to identify > > an *action* seems abusive, since the scheme obviously is meant to identify > > MMS destinations, just like the http scheme obviously is meant to identify > > HTTP resources, not "the action of retrieving some resource via HTTP." > > Granted, the mailto scheme with its subject and body parameters currently > > identifies some kind of actions, but I don't agree with this either. I > > think the scheme should only identify e-mail addresses. (As a side note, it > > should probably have been called "email" from the beginning.) > > > > > >>One other comment, I am not sure that "mms:" is appropriate for mobile > > > > phone devices. It either allows the automatic issuance of an mms message > > without acknowledgement, a big problems since MMS messages cost real money, > > or there is a required acknowledgement step, in which case most vendors want > > a consistent user experience, so why not simply use the composer as the > > acknowledgement vehicle, which would allow single button acknowledgement and > > review in the same step. Hence, we could have used mmsto: in the first > > place. Of course there may be other use cases that I am not taking into > > consideration. > > > > I agree. The "mms" vs "mmsto" distinction seems confusing at best. Again, > > it may just be that I don't see the right use cases for it. I think the > > purpose needs to be clarified. > > > > > > Regards, > > -- > > Daniel Brockman > > daniel@brockman.nu > > > > ----- Original Message ----- > > From: <Bennett.Marks@nokia.com> > > To: <uri@w3.org> > > Sent: Monday, September 29, 2003 9:02 PM > > Subject: IETF Draft on mmsto: > > > > > > > > Ted, > > > > I read the draft, ( > > http://www.ietf.org/internet-drafts/draft-wugofski-mms-uri-scheme-00.txt) > > and understand that you have patterned the mmsto: after mailto:. I have one > > concern. That is that on mobile phones the most common usage patterns for > > mms and sms do not exactly match the usage patterns that are modeled for > > email on the wired web. > > > > For the email scheme, far and away the most common usage pattern is the > > reply, and hence the email address is in the primary (mandatory) position > > before the "?". This is one common pattern for mobile, but the other equally > > common pattern for sms and mms is forwarding a piece of fixed information > > (i.e. a URL or image) to a new party. In this case the universal mode for > > this is to have the composer "find" the address in the address book. The > > current specification for mmsto: would require a default telephone # or > > email address (the reply scenario), and make it non-intuitive for the > > forwarding case, where I would pre-specify the "body" but not the address. > > It would require the user to explicitly overwrite a bogus address supplied > > with the scheme, and if not overwritten, the message would go back to the > > default location, definitely not the desired result. > > > > Is there some way to add the ability to "null out" the recipient address, > > which would force the composer to go through address assignment. Perhaps > > explicit instructions to a composing application if the destination address > > is specified as ""? > > > > One other comment, I am not sure that "mms:" is appropriate for mobile phone > > devices. It either allows the automatic issuance of an mms message without > > acknowledgement, a big problems since MMS messages cost real money, or there > > is a required acknowledgement step, in which case most vendors want a > > consistent user experience, so why not simply use the composer as the > > acknowledgement vehicle, which would allow single button acknowledgement and > > review in the same step. Hence, we could have used mmsto: in the first > > place. Of course there may be other use cases that I am not taking into > > consideration. > > > > > > > > Bennett Marks > > Vice Chair OMA Mobile Applications Environment > > Manager Infotainment Technology - MSW/SA > > Nokia Mobile Phones / Mobile Software Unit > > * 5 Wayside Rd., Burlington MA 01803 NOKIA > > * bennett.marks@nokia.com > > * +1 781 308 6556 > > * +1 781 993 1911 > > > > > > > > > > -- > > Ted Wugofski > CTO Office > Openwave > +1 817 658 6195 (m) > +1 817 737 4533 (o) > ted.wugofski@openwave.com > > >
Received on Tuesday, 30 September 2003 02:13:08 UTC