Discussion:
ReSpec and how it gets used
Shane McCarron
2016-05-11 19:31:32 UTC
Permalink
As one of the ReSpec maintainers, I am often concerned that changes we make
to the implementation will break stuff. And given some recent events, I
think that concern is merited.

One question that keeps coming up is "how is ReSpec used?" Is it only used
by document editors and the publication system, is it also used by the
working groups and reviewers (in that they are looking at live documents),
or is it used even more broadly?

We know that ReSpec is used by communities outside of the W3C, for example.

My basic concern can be characterized as "if we start using [insert new
cool browser feature] in ReSpec, will our constituents still be able to use
ReSpec?" Or, to turn that around, "Are there people using ancient user
agents actively accessing live documents written with ReSpec?"

So, a few questions for the group:

1. Do you have a feel for how ReSpec is being used?
2. Is there a policy about how far "back" we need to continue supporting
user agents?
3. Is there a way we can look at the W3C web server logs to get a feel
for what sorts of user agents are retrieving ReSpec from the W3C servers?
4. Would it be reasonable to provide some diagnostic information about
the user agent when we make a request to specref (so that we can do
analysis on usage)? Maybe only the first time a specific document is
requested (e.g., set a cookie)?
--
Shane McCarron
Projects Manager, Spec-Ops
Tobie Langel
2016-05-11 20:09:04 UTC
Permalink
1. Would it be reasonable to provide some diagnostic information
about the user agent when we make a request to specref (so that we
can do analysis on usage)?  Maybe only the first time a specific
document is requested (e.g., set a cookie)?
I don't really want to get into the business of gathering user data (not
to mention being compliant with EU cookie laws). I also don't think this
would help us very much to identify non-W3C scenarios, as those don't
necessarily rely on the Specref.

To answer your question more directly, I think you should cater to the
community you want to grow and not try to be everything to everyone.

--tobie
Michiel Bijl
2016-06-02 18:23:40 UTC
Permalink
At APG we only use to link to branches during calls, that’s pretty much it.

—Michiel
As one of the ReSpec maintainers, I am often concerned that changes we make to the implementation will break stuff. And given some recent events, I think that concern is merited.
One question that keeps coming up is "how is ReSpec used?" Is it only used by document editors and the publication system, is it also used by the working groups and reviewers (in that they are looking at live documents), or is it used even more broadly?
We know that ReSpec is used by communities outside of the W3C, for example.
My basic concern can be characterized as "if we start using [insert new cool browser feature] in ReSpec, will our constituents still be able to use ReSpec?" Or, to turn that around, "Are there people using ancient user agents actively accessing live documents written with ReSpec?"
Do you have a feel for how ReSpec is being used?
Is there a policy about how far "back" we need to continue supporting user agents?
Is there a way we can look at the W3C web server logs to get a feel for what sorts of user agents are retrieving ReSpec from the W3C servers?
Would it be reasonable to provide some diagnostic information about the user agent when we make a request to specref (so that we can do analysis on usage)? Maybe only the first time a specific document is requested (e.g., set a cookie)?
--
Shane McCarron
Projects Manager, Spec-Ops
Michiel Bijl
2016-06-02 18:25:32 UTC
Permalink
Which is needless to tell you Shane, as you are in it. Regardless, is your question whether it gets used outside of W3C?

—Michiel
Post by Michiel Bijl
At APG we only use to link to branches during calls, that’s pretty much it.
—Michiel
As one of the ReSpec maintainers, I am often concerned that changes we make to the implementation will break stuff. And given some recent events, I think that concern is merited.
One question that keeps coming up is "how is ReSpec used?" Is it only used by document editors and the publication system, is it also used by the working groups and reviewers (in that they are looking at live documents), or is it used even more broadly?
We know that ReSpec is used by communities outside of the W3C, for example.
My basic concern can be characterized as "if we start using [insert new cool browser feature] in ReSpec, will our constituents still be able to use ReSpec?" Or, to turn that around, "Are there people using ancient user agents actively accessing live documents written with ReSpec?"
Do you have a feel for how ReSpec is being used?
Is there a policy about how far "back" we need to continue supporting user agents?
Is there a way we can look at the W3C web server logs to get a feel for what sorts of user agents are retrieving ReSpec from the W3C servers?
Would it be reasonable to provide some diagnostic information about the user agent when we make a request to specref (so that we can do analysis on usage)? Maybe only the first time a specific document is requested (e.g., set a cookie)?
--
Shane McCarron
Projects Manager, Spec-Ops
Shane McCarron
2016-06-02 19:11:56 UTC
Permalink
APA / ARIA is an interesting case, in that we try to not expose the raw
ReSpec to the public. Since it is only "us" I am not super worried about
it. But there are some groups that make their ReSpec sources public and
encourage use. I was wondering how prevalent that is.
Post by Michiel Bijl
Which is needless to tell you Shane, as you are in it. Regardless, is your
question whether it gets used outside of W3C?
—Michiel
At APG we only use to link to branches during calls, that’s pretty much it.
—Michiel
As one of the ReSpec maintainers, I am often concerned that changes we
make to the implementation will break stuff. And given some recent events,
I think that concern is merited.
One question that keeps coming up is "how is ReSpec used?" Is it only
used by document editors and the publication system, is it also used by the
working groups and reviewers (in that they are looking at live documents),
or is it used even more broadly?
We know that ReSpec is used by communities outside of the W3C, for example.
My basic concern can be characterized as "if we start using [insert new
cool browser feature] in ReSpec, will our constituents still be able to use
ReSpec?" Or, to turn that around, "Are there people using ancient user
agents actively accessing live documents written with ReSpec?"
1. Do you have a feel for how ReSpec is being used?
2. Is there a policy about how far "back" we need to continue
supporting user agents?
3. Is there a way we can look at the W3C web server logs to get a feel
for what sorts of user agents are retrieving ReSpec from the W3C servers?
4. Would it be reasonable to provide some diagnostic information about
the user agent when we make a request to specref (so that we can do
analysis on usage)? Maybe only the first time a specific document is
requested (e.g., set a cookie)?
--
Shane McCarron
Projects Manager, Spec-Ops
--
Shane McCarron
Projects Manager, Spec-Ops
Marcos Caceres
2016-06-03 01:15:11 UTC
Permalink
Post by Shane McCarron
APA / ARIA is an interesting case, in that we try to not expose the raw
ReSpec to the public. Since it is only "us" I am not super worried about
it. But there are some groups that make their ReSpec sources public and
encourage use. I was wondering how prevalent that is.
Can you explain what "make their ReSpec sources public" means (or
point to some examples)? I'm not sure what that means exactly.
Shane McCarron
2016-06-03 13:19:28 UTC
Permalink
It means that the gh-pages branch is raw respec source, not static saved
output of respec. Examples? Sure.
http://w3c.github.io/webpayments-ig/VCTF/use-cases/ for example. Basically
everyone who uses respec at w3c EXCEPT ARIA / APA. As far as I know.
Post by Marcos Caceres
Post by Shane McCarron
APA / ARIA is an interesting case, in that we try to not expose the raw
ReSpec to the public. Since it is only "us" I am not super worried about
it. But there are some groups that make their ReSpec sources public and
encourage use. I was wondering how prevalent that is.
Can you explain what "make their ReSpec sources public" means (or
point to some examples)? I'm not sure what that means exactly.
--
Shane McCarron
Projects Manager, Spec-Ops
Markus Lanthaler
2016-06-05 12:15:05 UTC
Permalink
Post by Shane McCarron
It means that the gh-pages branch is raw respec source, not static
saved output of respec. Examples? Sure. http://w3c.github.io
/webpayments-ig/VCTF/use-cases/ for example. Basically everyone who
uses respec at w3c EXCEPT ARIA / APA. As far as I know.
We, the Hydra W3C Community Group also have dynamic Respec documents.
Recently I more often got reports that those documents don't render properly for some users. The last one I heard was that a fully patched IE11 can't render them.. I haven't verified that yet though.


Cheers,
Markus


--
Markus Lanthaler
@markuslanthaler
Shane McCarron
2016-06-05 14:10:53 UTC
Permalink
Can you send a link to the document? We can do some testing.
Post by Markus Lanthaler
Post by Shane McCarron
It means that the gh-pages branch is raw respec source, not static
saved output of respec. Examples? Sure. http://w3c.github.io
/webpayments-ig/VCTF/use-cases/ for example. Basically everyone who
uses respec at w3c EXCEPT ARIA / APA. As far as I know.
We, the Hydra W3C Community Group also have dynamic Respec documents.
Recently I more often got reports that those documents don't render
properly for some users. The last one I heard was that a fully patched IE11
can't render them.. I haven't verified that yet though.
Cheers,
Markus
--
Markus Lanthaler
@markuslanthaler
Markus Lanthaler
2016-06-05 15:48:16 UTC
Permalink
Post by Shane McCarron
Can you send a link to the document? We can do some testing.
The main one can be found at http://www.hydra-cg.com/spec/latest/core/

Thanks
Post by Shane McCarron
It means that the gh-pages branch is raw respec source, not static
saved output of respec. Examples? Sure. http://w3c.github.io
/webpayments-ig/VCTF/use-cases/ for example. Basically everyone who
uses respec at w3c EXCEPT ARIA / APA. As far as I know.
We, the Hydra W3C Community Group also have dynamic Respec documents.
Recently I more often got reports that those documents don't render properly for some users. The last one I heard was that a fully patched IE11 can't render them.. I haven't verified that yet though.


Cheers,
Markus


--
Markus Lanthaler
@markuslanthaler
Marcos Caceres
2016-06-07 02:40:29 UTC
Permalink
On June 5, 2016 at 10:15:12 PM, Markus Lanthaler
We, the Hydra W3C Community Group also have dynamic Respec documents. Recently I more
often got reports that those documents don't render properly for some users. The last
one I heard was that a fully patched IE11 can't render them.. I haven't verified that yet
though.
IE11 is not supported by ReSpec's generation mode, unfortunately (that
browser is over 2 years old now, and has been superseded by Edge).
Either kindly ask your users to switch to Edge or use a more modern
browser ... alternatively, please publish the ReSpec output instead,
which should work on any browser going back to IE6.
Travis Leithead
2016-06-07 15:00:37 UTC
Permalink
It's also possible your users were running into an issue recently patched for Edge (that also affected IE11): https://github.com/w3c/respec/pull/799

-----Original Message-----
From: Marcos Caceres [mailto:***@marcosc.com]
Sent: Monday, June 6, 2016 7:40 PM
To: Shane McCarron <***@spec-ops.io>; Markus Lanthaler <***@gmx.net>; spec-prod <spec-***@w3.org>
Cc: Michiel Bijl <***@agosto.nl>
Subject: RE: ReSpec and how it gets used

On June 5, 2016 at 10:15:12 PM, Markus Lanthaler
Post by Markus Lanthaler
We, the Hydra W3C Community Group also have dynamic Respec documents.
Recently I more often got reports that those documents don't render
properly for some users. The last one I heard was that a fully patched
IE11 can't render them.. I haven't verified that yet though.
IE11 is not supported by ReSpec's generation mode, unfortunately (that browser is over 2 years old now, and has been superseded by Edge).
Either kindly ask your users to switch to Edge or use a more modern browser ... alternatively, please publish the ReSpec output instead, which should work on any browse
Tab Atkins Jr.
2016-06-07 15:40:13 UTC
Permalink
Post by Marcos Caceres
On June 5, 2016 at 10:15:12 PM, Markus Lanthaler
We, the Hydra W3C Community Group also have dynamic Respec documents. Recently I more
often got reports that those documents don't render properly for some users. The last
one I heard was that a fully patched IE11 can't render them.. I haven't verified that yet
though.
IE11 is not supported by ReSpec's generation mode, unfortunately (that
browser is over 2 years old now, and has been superseded by Edge).
Either kindly ask your users to switch to Edge or use a more modern
browser ... alternatively, please publish the ReSpec output instead,
which should work on any browser going back to IE6.
As a reader of ReSpec'd specs, I'd highly appreciate it if more people
published the generated output instead of the sources. It avoids the
flash-of-unprocessed-content and subsequent anchor-jumping, and it
works better in the tooling infrastructure.

~TJ
Markus Lanthaler
2016-06-07 19:11:41 UTC
Permalink
Post by Tab Atkins Jr.
Post by Marcos Caceres
On June 5, 2016 at 10:15:12 PM, Markus Lanthaler
Post by Markus Lanthaler
We, the Hydra W3C Community Group also have dynamic Respec documents.
Recently I more often got reports that those documents don't render
properly for some users. The last one I heard was that a fully patched
IE11 can't render them.. I haven't verified that yet though.
IE11 is not supported by ReSpec's generation mode, unfortunately (that
Bummer
Post by Tab Atkins Jr.
Post by Marcos Caceres
browser is over 2 years old now, and has been superseded by Edge).
Either kindly ask your users to switch to Edge
Well, that might be tricky for lots of users as it also requires an OS update.
Post by Tab Atkins Jr.
Post by Marcos Caceres
or use a more modern browser ... alternatively, please publish the
ReSpec output instead, which should work on any browser going
back to IE6.
I'm not going to ask to start supporting IE11 again but what browser support do you aim for? Only the absolute latest version? IE11 still seems to have a considerable market share...
Post by Tab Atkins Jr.
As a reader of ReSpec'd specs, I'd highly appreciate it if more people
published the generated output instead of the sources. It avoids the
flash-of-unprocessed-content and subsequent anchor-jumping, and it
works better in the tooling infrastructure.
The thing I like most about ReSpec is that it doesn't need any "compilation" step. It's not perfect but for most use cases it works well enough.


--
Markus Lanthaler
@markuslanthaler
Shane McCarron
2016-06-07 19:36:47 UTC
Permalink
Comments in line
Post by Markus Lanthaler
Post by Tab Atkins Jr.
Post by Marcos Caceres
browser is over 2 years old now, and has been superseded by Edge).
Either kindly ask your users to switch to Edge
Well, that might be tricky for lots of users as it also requires an OS update.
Agreed. Moreover, this is not an option for many users (see below).
Post by Markus Lanthaler
Post by Tab Atkins Jr.
Post by Marcos Caceres
or use a more modern browser ... alternatively, please publish the
ReSpec output instead, which should work on any browser going
back to IE6.
I'm not going to ask to start supporting IE11 again but what browser
support do you aim for? Only the absolute latest version? IE11 still seems
to have a considerable market share...
We have no clear guidelines on this My inclination is to never break faith
with backward compatibility unless maintaining it forces a reduction in
primary function. I suspect we have made a mistake removing whatever
polyfill enabled IE11 support. And while I agree with some other
commenters that publishing static versions is a better end user experience
anyway, the reality is that many working groups are developing specs using
ReSpec, and those groups don't want to take the time to generate static
versions - in particular for their "Editor's Draft"s. So to the extent
that we want people to be able to readily review specs as they are in
development, we need to take this into consideration.

Note that I am not talking about the people writing the specs. I assume
they are working with relatively modern user agents. They are typically
geeks like us. But their constituents are often less tech-savvy. The Web
Payments community, for example, has A LOT of bankers in it. Conservative
organizations tend to lock down software and only upgrade rarely, and then
after acceptance testing. But they are nonetheless members of the W3C, and
should be able to review and comment on our specs.
Post by Markus Lanthaler
Post by Tab Atkins Jr.
As a reader of ReSpec'd specs, I'd highly appreciate it if more people
published the generated output instead of the sources. It avoids the
flash-of-unprocessed-content and subsequent anchor-jumping, and it
works better in the tooling infrastructure.
The thing I like most about ReSpec is that it doesn't need any
"compilation" step. It's not perfect but for most use cases it works well
enough.
Yes. It is what drew many of us to ReSpec. The recent instability has
made it a much less desirable platform. My groups have spent a lot of time
trying to resolve ReSpec introduced problems or learning new requirements
as features change. Worse yet, various changes have broken the tool chain
that enabled the automated generation of "TR" versions of specs or
otherwise made it impossible to publish without hand-editing.

As a maintainer of ReSpec, I am appalled. As a user, I am frustrated. As
an advocate, I am finding it a hard product to recommend.

Perhaps the solution is to make all the (named) versions available so that
document developers can choose the one that works well for them and their
users? Or identify a stable version and call that official, then leave the
"development" bleeding-edge that people can use or not. Do a migration to
stable periodically after substantial testing. I don't know. But something
needs to change. Right now I see the best case as people forking ReSpec so
they have something they can rely upon. I see the worst case as them
abandoning the platform. Both of these would be a failure.
--
Shane McCarron
Projects Manager, Spec-Ops
Marcos Caceres
2016-06-08 09:46:57 UTC
Permalink
Post by Shane McCarron
Comments in line
On Tue, Jun 7, 2016 at 2:11 PM, Markus Lanthaler
Post by Markus Lanthaler
Post by Tab Atkins Jr.
Post by Marcos Caceres
browser is over 2 years old now, and has been superseded by Edge).
Either kindly ask your users to switch to Edge
Well, that might be tricky for lots of users as it also requires an OS update.
Agreed. Moreover, this is not an option for many users (see below).
But this is an option for everyone who writes specs. I've never met
anyone at the w3c who is in this situation (or can't use another
browser).
Post by Shane McCarron
Post by Markus Lanthaler
Post by Tab Atkins Jr.
Post by Marcos Caceres
or use a more modern browser ... alternatively, please publish the
ReSpec output instead, which should work on any browser going
back to IE6.
I'm not going to ask to start supporting IE11 again but what browser
support do you aim for? Only the absolute latest version? IE11 still seems
to have a considerable market share...
We have no clear guidelines on this My inclination is to never break faith
with backward compatibility unless maintaining it forces a reduction in
primary function. I suspect we have made a mistake removing whatever
polyfill enabled IE11 support.
There was nothing removed. Just stuff got added. IE11 is not
maintained, so it breaks because it's been left behind by the Web.
Post by Shane McCarron
And while I agree with some other
commenters that publishing static versions is a better end user experience
anyway, the reality is that many working groups are developing specs using
ReSpec, and those groups don't want to take the time to generate static
versions - in particular for their "Editor's Draft"s. So to the extent
that we want people to be able to readily review specs as they are in
development, we need to take this into consideration.
Agree. This is a long process - but we need to work as a community to
get there.
Post by Shane McCarron
Note that I am not talking about the people writing the specs. I assume
they are working with relatively modern user agents. They are typically
geeks like us. But their constituents are often less tech-savvy. The Web
Payments community, for example, has A LOT of bankers in it. Conservative
organizations tend to lock down software and only upgrade rarely, and then
after acceptance testing. But they are nonetheless members of the W3C, and
should be able to review and comment on our specs.
Sure, and again the best way to serve them is to give them generated
snapshots (even of EDs).
Post by Shane McCarron
Post by Markus Lanthaler
Post by Tab Atkins Jr.
As a reader of ReSpec'd specs, I'd highly appreciate it if more people
published the generated output instead of the sources. It avoids the
flash-of-unprocessed-content and subsequent anchor-jumping, and it
works better in the tooling infrastructure.
The thing I like most about ReSpec is that it doesn't need any
"compilation" step. It's not perfect but for most use cases it works well
enough.
Yes. It is what drew many of us to ReSpec. The recent instability has
made it a much less desirable platform. My groups have spent a lot of time
trying to resolve ReSpec introduced problems or learning new requirements
as features change. Worse yet, various changes have broken the tool chain
that enabled the automated generation of "TR" versions of specs or
otherwise made it impossible to publish without hand-editing.
This is no different to Bikeshed or Anolis or any other piece of
software. Software breaks, things change. We patch stuff quickly and
move on.
¯\_(ツ)_/¯
Post by Shane McCarron
As a maintainer of ReSpec, I am appalled. As a user, I am frustrated. As
an advocate, I am finding it a hard product to recommend.
It's open source, you are free to leave, fork, use BikeShed, whatever
¯\_(ツ)_/¯.

I like the improvements I've made - and sure, there was a little pain
for a tiny number of people for maybe a couple of hours, but whatever.
At least it's actually getting maintained and updated now - and it's
vastly better.

A year ago, it was "stable" in the sense that it was rotting away
because no one was spending any time improving it after Robin left.
That's not stability: that's just bit rot.
Post by Shane McCarron
Perhaps the solution is to make all the (named) versions available so that
document developers can choose the one that works well for them and their
users? Or identify a stable version and call that official, then leave the
"development" bleeding-edge that people can use or not.
That's what we do today. We develop in branches, which go to
"develop", which then get released into "gh-pages".
Post by Shane McCarron
 Do a migration to
stable periodically after substantial testing. I don't know. But something
needs to change. Right now I see the best case as people forking ReSpec so
they have something they can rely upon. I see the worst case as them
abandoning the platform. Both of these would be a failure.
I think you are totally over dramatizing things. Little bugs are no
big deal. Most people haven't noticed that we've done like 40+
releases in the last year.
Matt King
2016-06-08 22:47:30 UTC
Permalink
I don't know if this is the right thread for this comment ... if not, feel free to let me know.

I am an editor and I rely on the JAWS screen reader. Because Firefox has to be updating is accessibility tree as respect runs, it takes a really long time to run. It is rare that I am able to start reading a branch in rawgit in under a minute. The ARIA spec takes up to 2 minutes before I can read it. Then, sometimes, like today, things are very broken. Today, none of the roles, states, or property sections have headings or permalinks. I don't know if that is due to a new respec bug or a failure of respect to run completely, or a defect in my spec text. Today, I know it is not a defect in my text because I haven't changed it since it last worked.

I am wondering if there is a better way for respect to work. Is there a way to make all the respec changes without doing it on the live DOM and then replace the entire DOM or something like that. Content hidden with display none is left out of the AX tree, so maybe the whole DOM could be hidden while the processing is occurring ... maybe not great for everyone, but at least you would all have an experience that is more like mine <smile>.

Matt King

-----Original Message-----
From: Marcos Caceres [mailto:***@marcosc.com]
Sent: Wednesday, June 8, 2016 2:47 AM
To: Shane McCarron <***@spec-ops.io>
Cc: Michiel Bijl <***@agosto.nl>; spec-prod <spec-***@w3.org>
Subject: Re: ReSpec and how it gets used
Post by Shane McCarron
Comments in line
On Tue, Jun 7, 2016 at 2:11 PM, Markus Lanthaler
Post by Markus Lanthaler
Post by Tab Atkins Jr.
Post by Marcos Caceres
browser is over 2 years old now, and has been superseded by Edge).
Either kindly ask your users to switch to Edge
Well, that might be tricky for lots of users as it also requires an OS update.
Agreed. Moreover, this is not an option for many users (see below).
But this is an option for everyone who writes specs. I've never met anyone at the w3c who is in this situation (or can't use another browser).
Post by Shane McCarron
Post by Markus Lanthaler
Post by Tab Atkins Jr.
Post by Marcos Caceres
or use a more modern browser ... alternatively, please publish
the ReSpec output instead, which should work on any browser going
back to IE6.
I'm not going to ask to start supporting IE11 again but what browser
support do you aim for? Only the absolute latest version? IE11 still
seems to have a considerable market share...
We have no clear guidelines on this My inclination is to never break
faith with backward compatibility unless maintaining it forces a
reduction in primary function. I suspect we have made a mistake
removing whatever polyfill enabled IE11 support.
There was nothing removed. Just stuff got added. IE11 is not maintained, so it breaks because it's been left behind by the Web.
Post by Shane McCarron
And while I agree with some other
commenters that publishing static versions is a better end user
experience anyway, the reality is that many working groups are
developing specs using ReSpec, and those groups don't want to take the
time to generate static versions - in particular for their "Editor's
Draft"s. So to the extent that we want people to be able to readily
review specs as they are in development, we need to take this into consideration.
Agree. This is a long process - but we need to work as a community to get there.
Post by Shane McCarron
Note that I am not talking about the people writing the specs. I
assume they are working with relatively modern user agents. They are
typically geeks like us. But their constituents are often less
tech-savvy. The Web Payments community, for example, has A LOT of
bankers in it. Conservative organizations tend to lock down software
and only upgrade rarely, and then after acceptance testing. But they
are nonetheless members of the W3C, and should be able to review and comment on our specs.
Sure, and again the best way to serve them is to give them generated snapshots (even of EDs).
Post by Shane McCarron
Post by Markus Lanthaler
Post by Tab Atkins Jr.
As a reader of ReSpec'd specs, I'd highly appreciate it if more
people published the generated output instead of the sources. It
avoids the flash-of-unprocessed-content and subsequent
anchor-jumping, and it works better in the tooling infrastructure.
The thing I like most about ReSpec is that it doesn't need any
"compilation" step. It's not perfect but for most use cases it works
well enough.
Yes. It is what drew many of us to ReSpec. The recent instability has
made it a much less desirable platform. My groups have spent a lot of
time trying to resolve ReSpec introduced problems or learning new
requirements as features change. Worse yet, various changes have
broken the tool chain that enabled the automated generation of "TR"
versions of specs or otherwise made it impossible to publish without hand-editing.
This is no different to Bikeshed or Anolis or any other piece of software. Software breaks, things change. We patch stuff quickly and move on.
¯\_(ツ)_/¯
Post by Shane McCarron
As a maintainer of ReSpec, I am appalled. As a user, I am frustrated.
As an advocate, I am finding it a hard product to recommend.
It's open source, you are free to leave, fork, use BikeShed, whatever
¯\_(ツ)_/¯.

I like the improvements I've made - and sure, there was a little pain for a tiny number of people for maybe a couple of hours, but whatever.
At least it's actually getting maintained and updated now - and it's vastly better.

A year ago, it was "stable" in the sense that it was rotting away because no one was spending any time improving it after Robin left.
That's not stability: that's just bit rot.
Post by Shane McCarron
Perhaps the solution is to make all the (named) versions available so
that document developers can choose the one that works well for them
and their users? Or identify a stable version and call that official,
then leave the "development" bleeding-edge that people can use or not.
That's what we do today. We develop in branches, which go to "develop", which then get released into "gh-pages".
Post by Shane McCarron
Do a migration to
stable periodically after substantial testing. I don't know. But
something needs to change. Right now I see the best case as people
forking ReSpec so they have something they can rely upon. I see the
worst case as them abandoning the platform. Both of these would be a failure.
I think you are totally over dramatizing things. Little bugs are no big deal. Most people haven't noticed that we've done like 40+ releases in the last year.
Shane McCarron
2016-06-08 23:27:57 UTC
Permalink
I was aware that you were having issues Matt. I was not aware they were
quite this dire. Can you send us / me the URI to your rawgit tree so I can
figure out what the respec issue is? As far as I know the ARIA trees were
fixed to work with the current ReSpec a couple of weeks ago.

As to your suggestions.... the approach I had wanted to take was to set
aria-busy to true on the body element until respec and any extensions were
complete. ReSpec added a Promise that should facilitate this. Do you
think this approach would work?
Post by Matt King
I don't know if this is the right thread for this comment ... if not, feel
free to let me know.
I am an editor and I rely on the JAWS screen reader. Because Firefox has
to be updating is accessibility tree as respect runs, it takes a really
long time to run. It is rare that I am able to start reading a branch in
rawgit in under a minute. The ARIA spec takes up to 2 minutes before I can
read it. Then, sometimes, like today, things are very broken. Today, none
of the roles, states, or property sections have headings or permalinks. I
don't know if that is due to a new respec bug or a failure of respect to
run completely, or a defect in my spec text. Today, I know it is not a
defect in my text because I haven't changed it since it last worked.
I am wondering if there is a better way for respect to work. Is there a
way to make all the respec changes without doing it on the live DOM and
then replace the entire DOM or something like that. Content hidden with
display none is left out of the AX tree, so maybe the whole DOM could be
hidden while the processing is occurring ... maybe not great for everyone,
but at least you would all have an experience that is more like mine
<smile>.
Matt King
-----Original Message-----
Sent: Wednesday, June 8, 2016 2:47 AM
Subject: Re: ReSpec and how it gets used
Post by Shane McCarron
Comments in line
On Tue, Jun 7, 2016 at 2:11 PM, Markus Lanthaler
Post by Markus Lanthaler
Post by Tab Atkins Jr.
Post by Marcos Caceres
browser is over 2 years old now, and has been superseded by Edge).
Either kindly ask your users to switch to Edge
Well, that might be tricky for lots of users as it also requires an OS update.
Agreed. Moreover, this is not an option for many users (see below).
But this is an option for everyone who writes specs. I've never met anyone
at the w3c who is in this situation (or can't use another browser).
Post by Shane McCarron
Post by Markus Lanthaler
Post by Tab Atkins Jr.
Post by Marcos Caceres
or use a more modern browser ... alternatively, please publish
the ReSpec output instead, which should work on any browser going
back to IE6.
I'm not going to ask to start supporting IE11 again but what browser
support do you aim for? Only the absolute latest version? IE11 still
seems to have a considerable market share...
We have no clear guidelines on this My inclination is to never break
faith with backward compatibility unless maintaining it forces a
reduction in primary function. I suspect we have made a mistake
removing whatever polyfill enabled IE11 support.
There was nothing removed. Just stuff got added. IE11 is not maintained,
so it breaks because it's been left behind by the Web.
Post by Shane McCarron
And while I agree with some other
commenters that publishing static versions is a better end user
experience anyway, the reality is that many working groups are
developing specs using ReSpec, and those groups don't want to take the
time to generate static versions - in particular for their "Editor's
Draft"s. So to the extent that we want people to be able to readily
review specs as they are in development, we need to take this into
consideration.
Agree. This is a long process - but we need to work as a community to get there.
Post by Shane McCarron
Note that I am not talking about the people writing the specs. I
assume they are working with relatively modern user agents. They are
typically geeks like us. But their constituents are often less
tech-savvy. The Web Payments community, for example, has A LOT of
bankers in it. Conservative organizations tend to lock down software
and only upgrade rarely, and then after acceptance testing. But they
are nonetheless members of the W3C, and should be able to review and
comment on our specs.
Sure, and again the best way to serve them is to give them generated
snapshots (even of EDs).
Post by Shane McCarron
Post by Markus Lanthaler
Post by Tab Atkins Jr.
As a reader of ReSpec'd specs, I'd highly appreciate it if more
people published the generated output instead of the sources. It
avoids the flash-of-unprocessed-content and subsequent
anchor-jumping, and it works better in the tooling infrastructure.
The thing I like most about ReSpec is that it doesn't need any
"compilation" step. It's not perfect but for most use cases it works
well enough.
Yes. It is what drew many of us to ReSpec. The recent instability has
made it a much less desirable platform. My groups have spent a lot of
time trying to resolve ReSpec introduced problems or learning new
requirements as features change. Worse yet, various changes have
broken the tool chain that enabled the automated generation of "TR"
versions of specs or otherwise made it impossible to publish without
hand-editing.
This is no different to Bikeshed or Anolis or any other piece of software.
Software breaks, things change. We patch stuff quickly and move on.
¯\_(ツ)_/¯
Post by Shane McCarron
As a maintainer of ReSpec, I am appalled. As a user, I am frustrated.
As an advocate, I am finding it a hard product to recommend.
It's open source, you are free to leave, fork, use BikeShed, whatever
¯\_(ツ)_/¯.
I like the improvements I've made - and sure, there was a little pain for
a tiny number of people for maybe a couple of hours, but whatever.
At least it's actually getting maintained and updated now - and it's vastly better.
A year ago, it was "stable" in the sense that it was rotting away because
no one was spending any time improving it after Robin left.
That's not stability: that's just bit rot.
Post by Shane McCarron
Perhaps the solution is to make all the (named) versions available so
that document developers can choose the one that works well for them
and their users? Or identify a stable version and call that official,
then leave the "development" bleeding-edge that people can use or not.
That's what we do today. We develop in branches, which go to "develop",
which then get released into "gh-pages".
Post by Shane McCarron
Do a migration to
stable periodically after substantial testing. I don't know. But
something needs to change. Right now I see the best case as people
forking ReSpec so they have something they can rely upon. I see the
worst case as them abandoning the platform. Both of these would be a
failure.
I think you are totally over dramatizing things. Little bugs are no big
deal. Most people haven't noticed that we've done like 40+ releases in the
last year.
--
Shane McCarron
Projects Manager, Spec-Ops
Marcos Caceres
2016-06-09 03:50:13 UTC
Permalink
Post by Matt King
I don't know if this is the right thread for this comment ... if not, feel free to let me know.
I am an editor and I rely on the JAWS screen reader. Because Firefox has to be updating is
accessibility tree as respect runs, it takes a really long time to run. It is rare that
I am able to start reading a branch in rawgit in under a minute. The ARIA spec takes up to
2 minutes before I can read it. Then, sometimes, like today, things are very broken.
Note that Firefox is severely broken wrt accessibility. I too rely
heavily on a screen-reader and have had to stop using Firefox
altogether (and yes, I work for Mozilla... that's how terrible the
situation is).
Post by Matt King
 Today,
none of the roles, states, or property sections have headings or permalinks. I don't
know if that is due to a new respec bug or a failure of respect to run completely, or a defect
in my spec text. Today, I know it is not a defect in my text because I haven't changed it since
it last worked.
Problems has been with the transition to "e10s" (multi-process) in
Firefox. We are working on it - but, realistically, I don't expect
things to get better for at least 1-2 years.
Post by Matt King
I am wondering if there is a better way for respect to work. Is there a way to make all the
respec changes without doing it on the live DOM and then replace the entire DOM or something
like that. Content hidden with display none is left out of the AX tree, so maybe the whole
DOM could be hidden while the processing is occurring ... maybe not great for everyone,
but at least you would all have an experience that is more like mine .
That's actually a great idea. Could certainly make all changes on a
different document, then replace the document fragments.

We do exactly the above for markdown parsing. I'll investigate!

Please follow up here:
https://github.com/w3c/respec/issues/818

Marcos Caceres
2016-06-08 09:23:52 UTC
Permalink
On June 8, 2016 at 5:11:57 AM, Markus Lanthaler
Post by Markus Lanthaler
I'm not going to ask to start supporting IE11 again but what browser support do you aim
for? Only the absolute latest version? IE11 still seems to have a considerable market
share...
Not amongst the tiny group of people who produce specs.

Anyone publishing source documents without going through the spec
generator is doing harm to their readers. It's really inexcusable
given that we have had Echidna for over a year and we worked so hard
to get the auto-publish process in place at the W3C.

Even if you don't want to use Echidna, you can run
`./tools/respec2html.js` locally when you check in your spec to have
it generated for you.

If people have any problems with generation, I'm happy to help them get set up!

As one of ReSpec's maintainers, and for the benefit of those that rely
on our specs, I would really like to see less ReSpec buttons (and
FOUC!) and more specs on /TR/.

Let's work towards making that super easy.
Marcos Caceres
2016-06-08 09:13:22 UTC
Permalink
Post by Tab Atkins Jr.
As a reader of ReSpec'd specs, I'd highly appreciate it if more people
published the generated output instead of the sources. It avoids the
flash-of-unprocessed-content and subsequent anchor-jumping, and it
works better in the tooling infrastructure.
Agree. Working towards that... part of this making Echidna easier to
use/integrate into projects (and to put an end to Editor's drafts).

I'm trying different things to kill the FOUC and stop the jumping
around. But will mean some breaking changes, so need to get some
groups to stop using ReSpec's crappy "events" model.
Continue reading on narkive:
Loading...