Topic on VisualEditor/Feedback

bug 49904 not on roadmap

17
Martijn Hoekstra (talkcontribs)

Bug 49904 is not on the roadmap. To me, the lack of subst: support is often a blocker for me.

TheDJ (talkcontribs)

I think the right question there, is what is it in your workflow that requires subst ? And for what reasons is that subst'ing then being applied in this workflow ?

Martijn Hoekstra (talkcontribs)

Mostly speedy deletion patrol on en.wiki. I often replace speedy deletion requests with PRODs which must be subst:ed (for a reason I'm not 100% sure of). I'm not sure why this is the right question. Is subst:'ing templates deprecated or discouraged?

TheDJ (talkcontribs)

Because VE is not wikitext and subst is an extreme form of wikitext. If there is something you need, then the developers want to understand why you need it. In certain cases, an alternative solution might be sought.

For instance, I can imagine that in this case that they would reimplement some of the annotations for templates, or even for subst itself, to make sure that something like this just happens automatically. In other cases, perhaps an editor is just looking for 'action favorites' or something which. They are not trying to reimplement wikitext, they are building an editor that supports what you need to do and that is backwards compatible with wikitext syntax.

Looking specifically at PROD, it indeed seems to be implemented as a templated action for a set of templates, that insert current information (and thus need to be replaced).

Martijn Hoekstra (talkcontribs)

This line of reasoning still somewhat troubles me. I don't need much convinving that Wikitext is a heaping pile of terrible, or that this is difficult, and, frankly, just wrong. In my mental model a template substitution is similar to a macro application (and happens at "save-time"/"compile time") where a transclusion is similar to a function application (and happens at "display-time"/"run time"). In the VE model "display time" comes before "save time", while in the wikitext model the situation is reversed. I understand that brings challenges with it as the available context is different. I truely understand the brokenness of being able to detect during parsing (which happens at save time in case of template substitution) to detect if you're in a "save time" context or a "display time" context and adjust output accordingly. I feel your pain when I have to think about unsubst, and deciding at "save time" you're not going to do the substitution and just return the wikitext, including the brackets. This is all crap.

When I think about implementations I feel your pain even more; when following the wikitext model, this would mean that the content is substituted as soon as it is entered, but from a usability perspective this is terrible, as there is no way to go back and change the substitution afterwards if you just insert the VE model of the parsed template (if that is even possible, node fragments ftl) - parsed in "save time mode" - right into the VE model being edited. (While I acknowledge the terrible, I could live with this as a first stab at it by the way).

But template substitution is a fairly often used feature - and if it is used fairly often, there are valid use-cases for it. At an intellectual level I understand the desire to know why it is wanted, and what those use cases are, but the bottom line is that you can bet on it there are such use cases, even if you don't yet know them, and they'll just need to get implemented, no matter how much they suck on a technical level. It also means that knowing and understanding the use case won't influence the implementation.

I realise my earlier response was curt, and this response still has some irk in it. The reason for this is that - although I haven't edited much over the past months - I think I'm one of the few people on en.wp who tries to do everything with VE primarily, first of because it can be an amazing product (it is when it works, which is most of the time), and secondly because when I'm doing admin tasks I run across a larger part of the long tail of features than most other tasks, and I can provide the team with feedback on what is working for me and what isn't. A response that I interpreted as "well, we'll be the judge if we find the work you're doing now in wikitext valuable enough to carry over to VE", while at the same time aiming to ultimately completely replace wikitext (again, the latter is a good thing). For me, that doesn't really feel inviting to keep providing that feedback. I hope you can see this perspective as well.

Nnemo (talkcontribs)

I am not aware of an ultimate goal of replacing entirely wikitext with visual editing.

Whatamidoing (WMF) (talkcontribs)

There is no plan to replace or remove wikitext editing.

Jdforrester (WMF) (talkcontribs)

That is and has always been the goal, yes; the intent is that users will never have to touch wikitext for any kind of mainline contribution, be that on articles, discussions, policy work, media improvement or anything else. (A very few issues will probably not be replaced any time soon, like interface messages – the MediaWiki: namespace – but that's so rarely edited that it's not likely even users around for five years will notice they can't edit it with VE.)

However, just because wikitext editing will be entirely replaced with visual editing and the wikitext editor won't be needed by the vast majority of users doesn't mean that we're going to remove it, don't worry.

Spinningspark (talkcontribs)

This is one of the many problems that could be very simply solved if only the VE developers would allow insertion of wikitext within the VE environment.

Jdforrester (WMF) (talkcontribs)

I don't understand. Could you please explain what problem you think that solves?

Nnemo (talkcontribs)

Users coming without JavaScript will probably keep having to edit wikitext. Blind users too.

Jdforrester (WMF) (talkcontribs)

That's not true; VisualEditor is already better for blind and partially-sighted users according to the feedback we've had, though I'm sure there's lots more improvements we could make. JavaScript-less users, however, are more of an issue, true.

Martijn Hoekstra (talkcontribs)

I'm getting mixed signals here.

Jdforrester (WMF) (talkcontribs)

The word "replace" is being used in two different ways, hence the confusion. Sorry! It would be better to say that VisualEditor "supplants" wikitext editing but we are not removing it.

Whatamidoing (WMF) (talkcontribs)

User:Martijn Hoekstra, thanks for this note. This may be fixed this Thursday (appearing at the Wikipedias the following Thursday).

Whatamidoing (WMF) (talkcontribs)

User:Martijn Hoekstra, subst'ing templates is working now.

While I'm happy to have you using VisualEditor (and telling me when you encounter problems!), have you considered using Twinkle for this work? It might save you a little time.

Martijn Hoekstra (talkcontribs)

Thanks, that's great to hear! The problem with Twinkle for this is that I like to notify the tagger personally, and twinkle is a little peculiar with edit summaries when replacing CSD with PROD

Reply to "bug 49904 not on roadmap"