Define the Term "Packaged Web Publication"

[prototypo]

The PWP spec will require a formal definition of "Packaged Web Publication".

[prototypo]

@iherman has suggested, "If I take a PWP, I unpack it on a web site (let that be localhost), thereby giving it a Web address, the resulting Web site MUST be a valid WP."

I agree with that, and suggest that it helps in the definition of a PWP. Perhaps something like this would work?

A Packaged Web Publication is a single information resource which may be unpacked to produce a Web Publication when published on the Web.

[iherman]

To add to this definition: I know that, for some usage, there may be packaged web publications that are not "made for the Web", whatever that means. I realize that this may mean some restrictions that would make the proposed definition unsuitable (for that purpose). My approach would be that this could be one differentiating factor between PWP and EPUB4 (after all, we haven not yet defined what the latter means either).

[HadrienGardeur]

I really don't like this concept of unpacking. If a PWP contains resources that are spread across multiple domains + HTTP headers, good luck "unpacking" that.

I would rather avoid any reference to unpacking and frankly the definition is the last place where I would include it.

[mattgarrish]

I would rather avoid any reference to unpacking

Yes, stating what it is without reference to what it might become might be less confusing, too. This is the product of all of one minute of thought, but perhaps something like:

A Web Publication that has been packaged into a single information resource, enabling it to be transported and shared independent of any specific address.

Obviously, that's rough, but I'd be looking in a definition for an explanation of what purpose it serves on its own.

[lrosenthol]

@mattgarrish replace 'shared' with "stored" and I'm good with it!

[iherman]

I am fine with the definition in the PWP doc.

However, eventually (not necessarily at this moment, ie, not necessarily for the FPWD) we should say something on the effect that the packaging, ie, the transformation for a WP to a PWP, is reversible in some way. Or it is reversible under certain constraints. Some sort of a reversibility, ie, reference to the WP that was at the origin of a WP is necessary, otherwise anything we say, for example, about affordances, about user agent behaviours, etc, becomes irrelevant for PWP...

[prototypo]

It would be nice to be able to identify (and to agree to):

1. How a PWP relates to WP
2. How WP relates to a PWP
3. How a PWP relates to the Web

@HadrienGardeur 's point relates to (2). My proposed definition covers (1). @mattgarrish 's proposal covers (1) and says specific things about (3). I don't object to the others' approaches so much as now think the definition needs to be more broad.

The "classic" way to handle these sort of discussions within W3C is to have the working group evaluate a series of proposals. If we can resolve some of them then we can use those resolutions to define what PWP means.

Here, then, are some proposals to lead us in that direction:

• PROPOSED: A PWP is a single information resource.
• PROPOSED: A PWP is a packaged version of the content comprising a WP.
• PROPOSED: A PWP need not be published on the Web.
• PROPOSED: A PWP may be published on the Web.
• PROPOSED: If a PWP is published on the Web, it becomes a WP.

I suggest that the WG vote on those proposals in one of next couple of telecons. I would vote +1 to all them since I drafted them ;-)

[TzviyaSiegman]

Thanks, @prototypo. I mostly agree with your list. I think it might be redundant to include both

PROPOSED: A PWP need not be published on the Web. PROPOSED: A PWP may be published on the Web.

MAY implies need not. So, I would prefer to include the latter statement. I do understand that explicitly stating need not makes a firm statement, and I'm okay leaving that in if others feel strongly about it.

[mattgarrish]

PROPOSED: If a PWP is published on the Web, it becomes a WP.

Do you mean "is unpacked and published on the Web" here? If put on a web server in its packaged form, it's still a PWP.

I don't disagree with the unpacking aspect in the definition, but it sounded odd to lead with it.

[BillKasdorf]

I apologize that I haven't been following every post but a PWP doesn't become a WP, it already is a WP. Bill K

Sent from my iPhone

On Nov 28, 2017, at 8:31 PM, Matt Garrish notifications@github.com<mailto:notifications@github.com> wrote:

PROPOSED: If a PWP is published on the Web, it becomes a WP.

Do you mean "is unpacked and published on the Web" here? If put on a web server in its packaged form, it's still a PWP.

I don't disagree with the unpacking aspect in the definition, but it sounded odd to lead with it.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/w3c/pwpub/issues/9#issuecomment-347722609, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AIYxNe4wtRWAD2CUj4PSEQVk56YmgpVqks5s7LP8gaJpZM4QrHcP.

[GarthConboy]

I think I understand what Bill's saying, and there is conceptual truth in it. But, the contents of a PWP may never have been published on the Web (a PWP created from whole cloth never on the Web), so we need to protect this use case, and the case where a PWP will be rendered by a Reading Systems without ever publishing it on the Web. Probably a job for MAYs.

[mattgarrish]

FWIW, WP avoids publication on the Web as a determination of existence, too. A WP is only defined as presentable using open web technologies, which allows for internet, intranet, localhost, whatever. Where PWPs might be published is probably not all that germane to their existence, either.

[prototypo]

Thanks for all the comments.

@BillKasdorf and @mattgarrish raise an interesting point that I was (poorly) trying to uncover: What is the difference between a PWP placed on a Web server an a single information resource and a PWP that is unpacked (whatever that comes to mean) then placed on a Web server as multiple information resources?

[iherman]

@mattgarrish:

FWIW, WP avoids publication on the Web as a determination of existence, too. A WP is only defined as presentable using open web technologies, which allows for internet, intranet, localhost, whatever. Where PWPs might be published is probably not all that germane to their existence, either.

This is of course true. But, I think, it is perfectly acceptable to use "PWP is published on the Web" with that level of sloppiness with an accompanying Note that makes it clear that what this means is that the content may be served and accessed via HTTP (which is the case for localhost or the intranet, too). Otherwise all our definitions become hopelessly convoluted.

@prototypo:

What is the difference between a PWP placed on a Web server an a single information resource and a PWP that is unpacked [...] then placed on a Web server as multiple information resources?

I think that, putting my uninformed FRBR terminology hat on (and I am sure that Bill or Tim will jump at me): if the PWP/WP has a (canonical) identifier as part of its information items (which is a SHOULD and not an MUST) then the PWP on the Web server, as well as the WP that represents the collection of information resources are two manifestations of the same work.

(I am a little bit wary of this discussion, it may end up in a long architectural discussion that may very well want to pick up with the TAG after our FPWD-s are published...)

[iherman]

@prototypo, you said

I suggest that the WG vote on those proposals in one of next couple of telecons. I would vote +1 to all them since I drafted them ;-)

because we are under a time pressure, I would actually see whether we can have a consensus via github (the thumbs up feature of github may come in handy!) to see if we can move on towards a FPWD publication with this. FPWD content can change later.

I voted 👍, but I also agree with the friendly amendment of @TzviyaSiegman.

[mattgarrish]

Note that makes it clear that what this means is that the content may be served and accessed via HTTP

But is this all that important to note? And why bring in http?

A PWP can be accessed by file protocol. It can be attached to an email. It can be shared by dropbox. Why does publishing or the web matter in particular? The point of packaging is to make the WP portable beyond the web. That's what really defines it in my mind.

[iherman]

On 29 Nov 2017, at 13:57, Matt Garrish <notifications@github.com mailto:notifications@github.com> wrote:

Note that makes it clear that what this means is that the content may be served and accessed via HTTP

But is this all that important to note? And why bring in http?

Note that I say 'may' above (I should have said 'MAY':-). Talking about HTTP is simply a more precise way of saying 'it is on the Web'. Many regard 'the Web' as being exclusively the World Wide Web but, technically speaking, there is no difference between that Web and an intranet or, simply, the files I access on my machine via localhost with a server running on my machine (something that I routinely do). A PWP can be accessed by file protocol. It can be attached to an email. It can be shared by dropbox. Why does publishing or the web matter in particular? The point of packaging is to make the WP portable beyond the web. That's what really defines it in my mind.

I am not sure I understand this question. "WP" means Web Publication. Yes, PWP means that it may be used beyond the Web, but what we are talking about is the relationship between that PWP and its content. We are not trying do define something fundamentally different, content-wise…

But I may completely misunderstand your point.

[HadrienGardeur]

I really don't like the idea of using the terms "publishing a PWP" when we're actually talking about unpacking a PWP.

I can publish a PWP by keeping the package as is, there's absolutely no need to distribute individual resources.

We need different terms for those two notions and frankly unpacking is the least important one.

[mattgarrish]

but what we are talking about is the relationship between that PWP and its content

Are we? I'm not talking about the WP that's been packaged. I don't disagree with this statement:

PROPOSED: A PWP is a packaged version of the content comprising a WP.

What I'm reacting to are these:

PROPOSED: A PWP need not be published on the Web. PROPOSED: A PWP may be published on the Web. PROPOSED: If a PWP is published on the Web, it becomes a WP.

What does publishing mean in these cases? Is it the simple act of putting the single information resource on a Web server? If so, my point is that the first two really don't need saying. We don't need to delve into all the places the PWP might go.

What I think you might be saying(?) is that you also hold the first two true for the WP that is packaged. I wouldn't disagree with that, but it's not the same question.

[iherman]

I agree with @HadrienGardeur and @mattgarrish that the term "publishing", in this context, is not the most judicious one. Let me put aside that wording issue for a moment. Yes, what I meant by this term (I cannot speak for @prototypo) is simply the act of putting an information resource on a Web server (that may include an intranet or only localhost server, too).

I guess my most important item for me is:

PROPOSED: A PWP is a packaged version of the content comprising a WP.

The question is, from a conformance point of view, what this means, ie, how a PWP relates to a WP. And the only way I can express that is that, conceptually, if I unpack this (whether onto the Web or on the fly by a reader), the unpacked content is a conformant WP. I am not saying it is always happening that way, I am not saying a system is required to do this. What I do not want is that the content of the PWP includes concepts, syntaxes, etc, that are in violation with the WP spec. For example, it will have a manifest of some sort in the PWP; that manifest MUST follow the same syntax and requirements that we define for a WP (it MAY have some extra items, but that is different). In other words, I would not want to see, under the heading of a PWP, a package that contains resources and structures that could not be interpreted properly by, say, a WP aware browser.

That is how I interpret this statement:

PROPOSED: If a PWP is published on the Web, it becomes a WP.

and the reason I am in favour of it. @mattgarrish, how would you express that in other ways?

---

@mattgarrish, I think you are right that these two statements:

PROPOSED: A PWP need not be published on the Web. PROPOSED: A PWP may be published on the Web.

may not say too much put together. Maybe it is fine to leave them out, because they only seem to muddle the waters.

[HadrienGardeur]

that manifest MUST follow the same syntax and requirements that we define for a WP (it MAY have some extra items, but that is different)

@iherman I don't think that's possible given the current direction of our discussions on WP. Several times, we've excluded PWP from such discussions, even though we knew that it would impact things.

The whole WP Address/entry page discussion is a good example: we simply won't have such a URL for PWPs that never existed as WPs in the first place.

[mattgarrish]

PROPOSED: If a PWP is published on the Web, it becomes a WP. @mattgarrish, how would you express that in other ways?

But then I don't think we disagree. What I said somewhere above is that shouldn't this say "unpack and publish" it becomes a WP. That seems to be what you're saying, too. If I just put the PWP on a web server, to me it's still just a PWP on a web server (even if conceptually it is just a different manifestation of a WP).

Would something like the following capture your concern:

Packaged Web Publication A Web Publication that has been packaged into a single information resource, enabling it to be transported and stored independent of any specific address or protocol. A Packaged Web Publication is typically constructed from a published Web Publication (i.e., one that has a specific URL and is accessible via HTTP), but this is not a requirement. Similarly, it is possible to unpack a Packaged Web Publication and publish it as a Web Publication, although there are practical limitations to this (e.g., re-publishing cross-domain resources will require access to all the domains).

(Sorry for jumping to definitions again, but it's easier for me to conceptualize.)

[iherman]

@HadrienGardeur we shall see. For me, there is a big difference between a PWP and an EPUB4, and we should keep those two separate. I am fine with EPUB4 being more lenient in some ways, and I am also fine with an EPUB4 that 'never existed as a WP in the first place'. I am not sure I would want that for a PWP.

[iherman]

@mattgarrish I like the definition you gave. Let us see what our fearless editor, ie, @prototypo, says...

[HadrienGardeur]

@iherman that's just another way you can draw the line. But if we have different requirements for EPUB 4 instead of PWP, it just means that EPUB 4 won't be a valid PWP.

That's why I'm not a fan at all of ignoring PWP/EPUB 4 from our WP discussions, this leads us to a place that we don't want to go.

[iherman]

@HadrienGardeur yes, indeed, that is another way to draw the line.

I do not dismiss your issues. If we have a clearer idea on what we (ie, the WG) want and have on both the PWP and WP sides, we may have to revisit some of the terms and information items to see how these two fit together and where the problems are. This may influence how we refine these in view of WP, PWP, and EPUB4. But I believe to have an overall view of where we want to go, and then go into the details, and make compromises where we have to take them, is a viable approach...

[BillKasdorf]

Agreed. And to clarify: I have always considered a PWP to be a type of WP (in other words, apart from the packaging it is in every sense a WP) whether or not it has actually been distributed in unpackaged form. It seems to me that if a publication purporting to be a PWP does not conform to WP apart from the packaging, then it’s not a PWP. Taken another way: if you unpackage a PWP it should be a WP without further alteration.

Bill Kasdorf

VP and Principal Consultant | Apex CoVantage

p:

734-904-6252 m: 734-904-6252

ISNI: http://isni.org/isni/0000000116490786 ORCiD: https://orcid.org/0000-0001-7002-4786https://orcid.org/0000-0001-7002-4786?lang=en

From: GarthConboy [mailto:notifications@github.com] Sent: Tuesday, November 28, 2017 8:52 PM To: w3c/pwpub Cc: Bill Kasdorf; Comment Subject: Re: [w3c/pwpub] Define the Term "Packaged Web Publication" (#9)

I think I understand what Bill's saying, and there is conceptual truth in it. But, the contents of a PWP may never have been published on the Web (a PWP created from whole cloth never on the Web), so we need to protect this use case, and the case where a PWP will be rendered by a Reading Systems without ever publishing it on the Web. Probably a job for MAYs.

— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/w3c/pwpub/issues/9#issuecomment-347726349, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AIYxNXgJpPO9jj2LazLd3qnoIleRzEwUks5s7LjMgaJpZM4QrHcP.

[GarthConboy]

I am okay with Matt's proposed definition and Ivan's comments. It likely moves some of my "never on the Web" concerns from PWP down to EPUB 4. But, after an initial negative reaction, perhaps I can get around to living with that. (at least at this juncture!)

[lrosenthol]

I'm with @HadrienGardeur and @BillKasdorf on this one. An EPUB4 MUST BE a valid PWP which MUST BE a valid WP.

As for @mattgarrish's suggestion, I can live with it though I agree that "publish" is the wrong word long term...

[BillKasdorf]

Re: Leonard’s

I'm with @HadrienGardeurhttps://github.com/hadriengardeur and @BillKasdorfhttps://github.com/billkasdorf on this one. An EPUB4 MUST BE a valid PWP which MUST BE a valid WP. . . . this also means a WP MAY BE a PWP which MAY BE an EPUB 4.

Bill Kasdorf

VP and Principal Consultant | Apex CoVantage

p:

734-904-6252 m: 734-904-6252

ISNI: http://isni.org/isni/0000000116490786 ORCiD: https://orcid.org/0000-0001-7002-4786https://orcid.org/0000-0001-7002-4786?lang=en

From: Leonard Rosenthol [mailto:notifications@github.com] Sent: Wednesday, November 29, 2017 1:22 PM To: w3c/pwpub Cc: Bill Kasdorf; Mention Subject: Re: [w3c/pwpub] Define the Term "Packaged Web Publication" (#9)

I'm with @HadrienGardeurhttps://github.com/hadriengardeur and @BillKasdorfhttps://github.com/billkasdorf on this one. An EPUB4 MUST BE a valid PWP which MUST BE a valid WP.

As for @mattgarrishhttps://github.com/mattgarrish's suggestion, I can live with it though I agree that "publish" is the wrong word long term...

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/w3c/pwpub/issues/9#issuecomment-347950084, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AIYxNUJtmLxMKn7r3LDZYiLdNlSr7Khaks5s7aDPgaJpZM4QrHcP.

[mattgarrish]

I agree that "publish" is the wrong word long term...

Perhaps we can strip it this way:

Packaged Web Publication A Web Publication that has been packaged into a single information resource, enabling it to be transported and stored independent of any specific address or protocol. The Web Publication that is packaged does not have to originate on the Web (i.e., have a specific URL that is accessible via HTTP); the only requirement is that it conform to Web Publications [WPUB]. Similarly, it is possible to unpack a Packaged Web Publication to create a Web Publication, but there are practical limitations to doing so (e.g., re-publishing cross-domain resources will require access to all the domains).

In the normative requirements we can definitively state that the WP MUST conform to [WP].

But I don't want to take all the fun away from @prototypo, so I'll try to shut up now ... :)

[BillKasdorf]

An enthusiastic +1!

Bill Kasdorf

VP and Principal Consultant | Apex CoVantage

p:

734-904-6252 m: 734-904-6252

ISNI: http://isni.org/isni/0000000116490786 ORCiD: https://orcid.org/0000-0001-7002-4786https://orcid.org/0000-0001-7002-4786?lang=en

From: Matt Garrish [mailto:notifications@github.com] Sent: Wednesday, November 29, 2017 1:48 PM To: w3c/pwpub Cc: Bill Kasdorf; Mention Subject: Re: [w3c/pwpub] Define the Term "Packaged Web Publication" (#9)

I agree that "publish" is the wrong word long term...

Perhaps we can strip it this way:

Packaged Web Publication A Web Publication that has been packaged into a single information resource, enabling it to be transported and stored independent of any specific address or protocol. The Web Publication that is packaged does not have to originate on the Web (i.e., have a specific URL that is accessible via HTTP); the only requirement is that it conform to Web Publications [WPUB]. Similarly, it is possible to unpack a Packaged Web Publication to create a Web Publication, but there are practical limitations to doing so (e.g., re-publishing cross-domain resources will require access to all the domains).

In the normative requirements we can definitively state that the WP MUST conform to [WP].

But I don't want to take all the fun away from @prototypohttps://github.com/prototypo, so I'll try to shut up now ... :)

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/w3c/pwpub/issues/9#issuecomment-347957676, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AIYxNQp1wCKSJNVxAxIrYGBlNdM2IRulks5s7abzgaJpZM4QrHcP.

[prototypo]

Thanks for a productive conversation, everyone.

@mattgarrish Don't worry, there is plenty of fun left for everyone :)

[css-meeting-bot]

The Working Group just discussed Rerviewing/closing PWP issues, and agreed to the following resolutions:

• RESOLVED: accept the comment above as the definition for PWP for now

The full IRC log of that discussion

<ivan> Topic: Rerviewing/closing PWP issues
<tzviya> https://github.com/w3c/pwpub/issues/9
<dauwhe> Github: https://github.com/w3c/pwpub/issues/9
<NickRuffilo> ...: The two big issues remaining are issue 6 and 9. If you look through here - there's a comment where Dave Wood proposes some working definitions. Proposal: 'A PWP is a single resource...' ... There was some nit-picking, but it seems like we agree on a majority of the items. Did we have a final resolution?
<ivan> +1
<ivan> q+
<garth> https://github.com/w3c/pwpub/issues/9#issuecomment-347881185
<NickRuffilo> ...: We wanted to combine 'may be published on the web' and 'need not be published on the web' into one item.
<NickRuffilo> Garth: I thought that what Matt proposed was what people were agreeing on. It's similar to what you read.
<ivan> Proposed resolution: "A Web Publication that has been packaged into a single information resource, enabling it to be transported and stored independent of any specific address or protocol.
<ivan> A Packaged Web Publication is typically constructed from a published Web Publication (i.e., one that has a specific URL and is accessible via HTTP), but this is not a requirement. Similarly, it is possible to unpack a Packaged Web Publication and publish it as a Web Publication, although there are practical limitations to this (e.g., re-publishing cross-domain resources will require access to all the domains)."
<ivan> s/A Web Publication/A Packaged Web Publication/
<NickRuffilo> Garth: In the comment pasted above - it has web publication defined as (you can read comment above)
<garth> q?
<ivan> ack ivan
<NickRuffilo> Tzviya: On github people have mostly agreed on this - but lets check with all - is this what we're agreeing on?
<NickRuffilo> Jeff: One of the obvious things is that all resources are available to the package.
<ivan> q+
<NickRuffilo> Garth: There's a concept of a spine - and then secondary resources, the ones that are part of the publication. The primary and secondary would be in the Packaged part, but there can still be external resources. Such as a link to Wikipedia - which wouldn't be included.
<tzviya> ack ivan
<NickRuffilo> Ivan: The other typical example Jeff is referring to a specific font. The situation is that - in some cases you do not wish to include the font. The reader might provide a fallback, but that is an example of a resource that isn't part of the package but is used.
<NickRuffilo> Jeff: OK - this is a central issue in the work I'm currently doing. We work with lots of clients who have video, which breaks the epub3 spec. PDFs or other resources that are outside of the epub3 spec. So it's of real interest to me.
<NickRuffilo> Garth: Epub3 allows video resources to be external...
<NickRuffilo> Jeff: You are right, but PDF gives you errors
<garth> : https://github.com/w3c/pwpub/issues/9#issuecomment-347881185
<NickRuffilo> Garth: So, is this a fine definition? Any comments?
<garth> q?
<ivan> PROPOSED: accept the comment above as the definition for PWP for now
<jbuehler> Thank you Nick.
<garth> +1
<ivan> +1
<tzviya> +1
<NickRuffilo> +1
<wolfgang> +1
<garth> https://github.com/w3c/pwpub/issues/6
<ivan> RESOLVED: accept the comment above as the definition for PWP for now

[iherman]

Based on the discussion (see https://github.com/w3c/pwpub/issues/9#issuecomment-349037428), closing the issue.