Mark globbing as "at risk"

[RubenVerborgh]

As requested by @timbl: point in the spec to the discussion at #148 and #151.

[michielbdejong]

👍

[kjetilk]

Looks very timely.

[melvincarvalho]

Could, at a minimum, specs be versioned, please?

EDIT : It would appear I am unable to weigh in further on this thread. So apology in advance, this message appears out of sequence. While not an official W3C REC, solid to date has had some of the highest spec hygiene that I've seen anywhere. This new approach to standardization does not serve anyone well. Once against apologies that this message was out of sequence, I cant explain why I was unable to bottom post a message. I am therefore, unable to easily respond to the specific points raised. Fun times.

EDIT2 : Caught up with Ruben out of band. So I'm happy to go with Tim's version of "at risk", meaning this will go away in the long term, but will work for now. Which is exactly what I had in my head. My concern is just that time is allowed to go through different items raised. So I think ... all good.

EDIT3 : When I set my review to approve, it says "user is blocked". Wnen I dismiss it says "something went wrong". Must be in a strange state. So if you have to go ahead with this then do, versioning would be a nice to have.

EDIT4 : @Mitzi-Laszlo as observed by the image above, may I possibly ask you to investigate why my participation on this thread has been apparently limited? (see image above). My view on this may be unpopular but that does not mean it's wrong. I do also have the most experience here (12+ years) in building solid type apps, and also the most experience (5+ years) in implementing globbing. It's impossible to make progress on a topic if participation is not on an equal basis. Additionally I am blocked from updating the state of my review. Thanks alot in advance if you are able to offer some of your time. Issue should be on hold until it is working for everyone.

[melvincarvalho]

Would prefer this not to happen, as it's in use. The arguments made have not been throught through and there are several stake holders, which have yet to be fully contacted, that may wish to comment on this. It will take a ton of time to dissect the multitudinous false assumptions associated with this issue, and there has been little to no due diligence.

Basically, I dont see the need or the pressing need.

On a meta level. I would consider very carefully the precident set here. A number of people that do not use, or possibly even understand, a feature are in contention with a number of people that do. Yet there there is a heavy handed approach where unnecessary bike shedding changes are rail roaded through on a ridiculous 1 day time line. Consider someone less friendly to the project. They may remove something a competitor is using (actually that's the case here), and thre is no recourse. They may go on vacation and find their work obsolete overnight. No developer wants to build on quick sand. Or as a great man once said, this is no way to run a restaurant.

But if it has to happen, could, at a minimum, specs be versioned, please? This is the change that I request to this document, something similar to semver, but pick a system you prefer. Whether that filters up to parent specs and change log is up to you.

[RubenVerborgh]

The point of this is that stakeholders see it, no? We explicitly invite them to the relevant issues. “At risk” (to me) means: make yourself heard if you want to keep it. Other suggestions to attract stakeholders welcome.

[melvincarvalho]

Other suggestions to attract stakeholders welcome.

• identify our apps that use globbing (eg in our main 2 orgs)
• identify the main authors (e.g. for cimba which was both software and a paper)
• reach out to them with your proposed alternative
• give a time line on which you intend to change status (the last one you gave was 1 day, which should scare the living daylights anyone sane, I hope something more sensible can be proposed)

“At risk” (to me) means: make yourself heard if you want to keep it.

Had no idea about this, as you have had 4 different views in fewer days, so that's good to know.

[RubenVerborgh]

as you have had 4 different views in fewer days

I fail to see why that was necessary.
I have only one view: globbing must go.

The other PR was to correct errors in the spec you pointed out.
The project I created was to give you a client-side alternative to globbing.
This PR is on request by Tim so others can locate these issues.

[TallTed]

@RubenVerborgh @melvincarvalho @kjetilk @michielbdejong (@timbl) -

I think it important to remember that a publicly posted and available spec in a sphere of "permissionless development," especially a spec developed outside of any spec developing/vetting/blessing organization (so far as I'm aware, the Solid spec did not go through W3, IETF, nor other similar processes), does not offer a good way to reach to all stakeholders who may be impacted by changes to (nor who might offer good input to the content of) that spec.

That said, this spec does not strike me as nearly the same sort of creature as an IETF RFC or a W3 Rec. Solid server apps and client apps based on this spec are rather fewer than those based on SMTP, NNTP, HTTP, etc., so the number of possible implementation stakeholders is likewise rather smaller.

Also, given that the Solid spec remains in a 0.x version (the only version I see is 0.5.0, a version generally understood to indicate still under development), it seems to me that everything is "at risk", at least until a 1.0 is posted, and that there's a degree of "let the buyer beware" inherent in any implementation based on such a 0.x spec .

[elf-pavlik]

This PR is on request by Tim so others can locate these issues.

@melvincarvalho as I try to follow globing related threads everyone who currently actively participates in the project seems to already voiced their opinion. I think everyone had preference to remove that feature, only you seem willing to defend it. I think this gives a ground to mark this feature as 'at risk' and link to related issues, giving this way people who read the spec, possibly wanting to implement it, chance to weigh in.

• identify our apps that use globbing (eg in our main 2 orgs)
• identify the main authors (e.g. for cimba which was both software and a paper)
• reach out to them with your proposed alternative

Since everyone else who expressed their opinion gave 👍 to remove globing, if you care about this feature and would like others to change their stand, you may need to take the steps you suggested yourself. While this PR IMO should land ASAP, #151 can stay pending as long as you want to actively work on changing current rough consensus on removing this feature.

timbl's comments have removed concerns about an abrupt change of state