Want to take part in these discussions? Sign in if you have an account, or apply for one below
Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.
Hey! I think we have our first decision making test!
It relates to matters of style.
I contribute technical content to the nLab whenever I can, but most of what I can offer comes in the form of "beautification". This is helpful to me because I learn as I edit pages.
I've been attempting to replace ascii links with symbolic links and place redirects to the ascii pages. For example -categories gets redirected to infinity-category, etc.
I've also been adding redirects for plural links. For example, categories is now redirected to category.
Zoran recently modified Drinfel'd twist where I noticed two links that had the undesirable (to me anyway) plural links bialgebra cocycles and Hopf algebras.
I changed these to bialgebra cocycles and Hopf algebras and added the appropriate redirects.
Zoran was not happy with this and changed the links BACK to bialgebra cocycles and Hopf algebras.
Should we let this slide? Should we decide on matters of style like this and enforce them?
Based on Toby's comment on the Latest Changes page, it seems like he will continue to change "]]s" to "s]]" when he comes across them EXCEPT for modifications made by Zoran. Personally, I am not interested in checking to see who made the change and am inclined to go ahead and change "]]s" to "s]]" while adding the necessary redirects. I don't want to do this if Zoran is going to go back and change things back to "]]s".
In the grand scheme of things, this is not very important, but I'd like to have an n-community decision made as to what to do about it since there does seem to be a conflict.
How do we resolve this?
My vote is that we develop a "Matters of Style" page that has some guidelines regarding issues like this.
On it, I would probably add something stating that we should not change "s]]" to "]]s" and that "s]]" is the preferred form for plural links.
I'm not going to spend a lot of time checking who made an edit; mostly I'll just consider whether the page is one likely to have been written by Zoran or not. I'll guess and make some mistakes, which is not really a big deal. I will certainly be able to avoid changing things in a page that Zoran just created; that will be easy! … Actually, it really works the other way around. Links in a ]]s
style written by most people exist only because we didn't have redirects once; I intend only to change those. Eventually it will be appropriate to presume that anything written that way was written that way on purpose, and then I won't change anything.
I absolutely do not think that we should have a Matters of Style page. This just creates unnecessary conflicts when somebody learns that the MoS insists on doing things the wrong, even stupid, way (in their opinion, of course). Let people write things the way that they want to write them! Of course, nobody owns a page, and eventually things will probably end up in the majority style, but it's frustrating to see that, after you write a nice page, somebody has edited it just to make things (slightly, and in your opinion if not theirs) worse. (In contrast, if they improve the content when they worsen the style, then that is a clear improvement overall and not so frustrating.)
Nobody has been messing with my apostrophes, we have both single and double quotation marks (not to mention sentential punctuation both inside and outside of the quotation marks), British and American spelling conventions (except in page names), … these haven't caused any conflicts, because nobody has been going around trying to change things one way and upsetting the people that prefer them the other way. Sure, if I were in charge, then I could teach you all about the difference between an apostrophe and a quotation mark, the logical status of a quotation, the futility of spelling reform, … but even though I've made more edits to more pages than anybody else, I've never tried to provoke conflicts on these matters and I won't.
I have noticed that you (Eric) sometimes try to change my preferred $\infty$-category
to your preferred ∞-category
. If that's all that you did, then I might pull a Zoran and change it back. But I wouldn't touch it if you were simply the first person to fix an original infinity-category
, even though it does produce the wrong font (^_^).
Yeah. I stopped changing things when it looks like it is written for itex fonts intentionally. Especially for thing like L-infinity, A-infinity, E-infinity, where the infinity is a subscript.
However, when infinity is not a subscript, I don't see any big difference. Even then, I won't change itex back to unicode, but I probably will never change ascii to itex, since I prefer the unicode infinity links.
After thinking about it, I kind of agree on the whole MoS thing. I mostly change the links to plural to intentionally "break" them. Then I can skim the page for broken links and add redirects where needed to fix them. At that point (once the redirects have been added), if someone wants to change the links back, then no harm is done.
I agree with Toby here. Taken to its logical extreme, then we could have someone going round changing all my British spelling into American and then where would we be?
Isn't an easier way of finding the required redirects to go to the "All pages" page and scan down the "wanted by" list? Ah, I think I see - some of these aren't actually wanted redirects yet but will become wanted when someone changes category to just categories.
Mind you, the quickest way is to export the whole lot as Markdown (which, I believe, can still be done) and then do
grep -rh '\[\[[^]]*|[^]]*\]\]' *
That ought to find you all the wiki-links that are pretending to be something that they ain't.
grep -rh '\[\[[^]]*|[^]]*\]\]' *
Gosh, that's a lot! (^_^)
<p>Hi Toby,</p>
<p>A systematic process I've been going through (which could be changed if necessary) is to</p>
<ol>
<li>Change all instances of "infinity" and "omega" (but not $\infty$ and $\omega$) to unicode <img src="/extensions/vLaTeX/cache/latex_ff52bcaf24f3d4af5a2a50cf5200b74f.png" title="\infty" style="vertical-align:-20%;" class="tex" alt="\infty" /> and <img src="/extensions/vLaTeX/cache/latex_b14ea9be9b8af885841ec254915f1561.png" title="\omega" style="vertical-align:-20%;" class="tex" alt="\omega" /> respectively</li>
<li>Change all instances of "]]s" and "]]es" to "s]]" and "es]]" respectively</li>
<li>Scan for all broken links and add the necessary redirects to fix them</lI>
</ol>
<p>Now I see that you are changing links from "s]]" back to "]]s" and even creating pages with links containing "]]s".</p>
<p>Unless I change my system, then I will invariably end up changing links that you've created and maybe even links that you've changed from "s]]" to "]]s".</p>
<p>[Edit: I see it is worse than that. On homotopy group, the only change you made was to change two links back from "s]]" to "]]s" after I just changed them from "]]s" to "s]]" (???)]</p>
<p>Should I change my system?</p>
<p>I like my system because it reveals (with as little thought as possible) pages that could use redirects.</p>
<p>What is the motivation for creating links with "]]s" as opposed to "s]]" and letting redirects take care of it?</p>
<p>To me, and maybe it is just me, a link with "]]s" just looks odd and I instinctively want to change it.</p>
<p>This is one reason why a "Matters of Style" page might be helpful. It doesn't have to be a hard set of rules, but at least having some guidelines so that we avoid wasted effort would be good I think. For example, if a Matters of Style page said that we prefer "]]s" to "s]]" then I would never again change a link from "]]s" to "s]]".</p>
<p>Again, the reason I am changing these links is to intentionally break them in order to reveal places where a redirect might be useful. I try to never leave a link broken.</p>
<p>Thanks</p>
There are two situations in which I've been changing s]]
back to ]]s
:
]]s
exists and you never created a redirect for s]]
;]]s
nor s]]
exists.At homotopy group, I started to do (1), but in the end I created the missing redirect instead; usually I'm too lazy to do that. I assume that (at least when it sits around for a couple of days) this is a mistake on your part when this happens.
The reason that I change back (2) is not because of any MoS-like preference for ]]s
, but rather the Naming Convention preference for singular names. (What I'm worried about is somebody, completely ignorant or just not caring about the issues surrounding plural names and redirects, clicking on the link to create the page. I don't want pages created under both names, so I try to have only one name linked when the page doesn't exist, and that name is the singular.) I would not change back …s
(but again I've been too lazy to make that myself).
So the upshot is that I change back any s]]
link that doesn't actually work, for one reason or another. (I wouldn't change it back if you just created it, since then I assume that you're still in the process of going through and searching for redirects to create. But homotopy group was two days old!)
Maybe we should compromise: If you'll agree not to create new plural links to pages that don't exist (either creating a redirect, as you have been, or making a …s
link when no page exists yet to redirect to), then I'll agree not to simply change back bad plural links that I see but instead take the time to do something more intelligent with them (either creating the redirect myself or making a …s
link when no page exists yet to redirect to).
As for writing new ]]s
links, I do that when I don't think that the s]]
redirect exists yet and I'm too lazy to make one. I probably won't stop that, but I certainly don't object (like Zoran would) if you change those and (if necessary) create the redirect.
I know you said ignore your previous statement, but those were not technically broken links. The singular versions of the links did not exist yet so I could not add a redirect. The last time that happened, I created a blank version of the singular link JUST so I could add the redirect and unbreak the link and you didn't seem to like that either, so I was just trying to make you happy :)
Ok. Now I will wait to see what you have to say :)
At least now I see what made you want to "fix" the ink. It makes sense, but like I said, my system will never leave a broken link UNLESS the page I'd like to add a redirect to does not exist yet (which should be pretty rare).
To ensure that my system never leaves a "broken" link, I could create a blank page if it doesn't exist and add the redirect. That way, when someone DOES add content, the redirects will work automagically.
Anyway, I am open to changing my sytem, but I do hope to have some kind of system.
And yeah, I don't like the blank pages with nothing but a redirect either; I find them very confusing! (Not when I reach them, of course, but before I reach them. I expect there to be something when I get there!)
Ok!
I think we are getting somewhere.
I will modify my system so that it should never (human error aside) result in broken links. In fact, the current system should already not generate any broken links, but possibly "wanted" links. For example, on homotopy group I changed homology groups to homology groups. Since homology group did not exist yet (it was wanted), we ended up with a "wanted" homology groups as well. I didn't consider this to be a "broken" link. Just "wanted".
Since you considered this to be "broken", you changed it back to homology groups, which didn't change the fact that it was still wanted.
From now, if page does not yet exist, I will not change a link to pages resulting in two wanted pages. Instead, I will change the link to pages. This way, only one wanted page remains.
Otherwise, in general, when I come across pages (except maybe something obviously done by Zoran), I will change it to pages and add the redirect if needed.
Now, to avoid the situation where I change a link, you change it back, then I end up changing it AGAIN, I'd ask you to never change a link from pages to pages. If you happen to come across a legitimately broken or wanted page and feel the desire to "fix" it, could you also change it to pages?
Between you and me, I think we account for 95% (or more) of all cosmetic changes to the nLab, so if we can agree on something, we should never end up doing wasteful changes.
On a related note, I think that we should consider changing the outdated singular name preference. I think that is appropriate in a world without redirects, but now is no longer needed. I think people should feel free (in fact encouraged) to use the plural name when appropriate. In time, all pages will have all obvious redirects, so the chance of anyone accidentally creating a page with the plural name will become increasingly remote. In the unlikely event that it does happen, it is very likely that someone (probably your or me) will see it, and quickly make the corrections and add the missing redirects.
Once we have systematically created a majority of likely redirects, we can (and I intend to) begin working through all "Wanted Pages". At that point, if you write new content including a link to pages when page exists and does not yet have a redirect, then I (or anyone else interested in beautifying the nLab) will see it in the "Wanted Pages" and can add the redirect.
I really don't think anyone except Zoran would be offended if I immediately changed any links in any new content they create to the plural versions. So unless told otherwise, I will probably continue to do that.
I agree that one should (normally) say ‘wanted’ instead of ‘broken’. In those terms, my problem is that homology groups is not really what we want; we want homology group instead. So homology groups, which appears to be a wanted link, wants the wrong thing; thus it is broken!
But that is just terminology. As far as the process, we are agreed … at least on everything before ‘On a related note’.
It's true that naming conventions are not so important in a world without redirects, but I still think that they're a good idea, for same the reasons that make them useful on Wikipedia (which also has redirects). Basically, if wanted links and new pages are always (or usually) made following such conventions, then all of the wanted links will work as soon as the new page is created, even if someone like you or I does not keep track of everything and fill in appropriate redirects. Someday (probably late August or September, if I get a teaching job, but someday in any case), the Lab will be too large for me to follow it all; probably it will be be too large for you to follow it all, someday. Then we will be happy if people like Urs have certain standard naming conventions (whatever those conventions are) ingrained in them (as Urs by now has, I think).
Oh right. I realize I didn't say exactly what I meant to say.
Of course I think we should have a naming convention for the pages themselves, but what I meant is that I don't think the name convention for the page should imply a name convention for links. For example, I think it makes sense to have page names be singular lower case nouns as we've been doing it, but I don't think this should imply a preference for pages over pages.
So in my "On a related note", whenever I said "name", it should be read "named link".
I think we're pretty much on the same page now.
I agree, no convention for links in general. But the naming convention for pages should apply to wanted links, since these give the names that people will create for the pages.
Toby wrote:
Gosh, that's a lot! (^_^)
Just out of interest, is your remark in response to running the script and seeing its output or something else? Incidentally, if the list initially appears long it could just be lots of repetitions. Piping the output through sort
and then uniq
might cut it down a lot.
Trying to make everything look the same is like a broken pencil. I think that we should be careful, when advocating certain styles, that we don't make it difficult for anyone to use or contribute to the wiki. Thus ensuring that non-existent links point directly to the page that ought to be created is a Good Thing. Thus apples is good if "apple" does not yet exist. I would therefore be extremely reluctant to change this.
Changing stuff to see what needs doing is a misuse of tools. It's using a screwdriver to bang in a nail. There are other ways of finding out what needs doing, which don't involve altering pages (and thus potentially annoying other users and stacking up the revision numbers). I've given you some code for finding the links
. I could also offer a script to sort out the list of wanted pages from redirected pages in the list
offered by n-lab.
Note: I'm not suggesting that we don't send a feature request to Jacques that this list be more sensibly organised to take redirects into account. That would obviously be a very useful change, and is a reasonable one to suggest. Rather, I'm saying that when doing "meta" stuff like finding out lists of things, one ought to have a "non-interventionist" stance wherein one tries to find a way of doing it without actually altering pages.
I suppose that having weighed in, I ought to say something about the original matter. The case for pages over pages seems obvious, especially when one applies it to something like categories over categorys. However, I would caution being careful changing content without considering the context. Since links have a visual display then they can be used to add emphasis, or to indicate which part of a word or phrase is the core piece. Moreover, if there is ambiguity then the syntax
can be a way out. For example, almost everyone called their version of a smooth space a "differentiable space". If I want to quote a result using, say, Chen spaces then I have a choice: use the original terminology, or modify it. Neither is completely satisfactory, but if I can write differentiable space
then I'm in a win-win situation.
Just out of interest, is your remark in response to […]
I found my latest download of the Markup export, unzipped it, and ran your grep
script on it. Yes, one might extract just the links and run sort -u
(no need for uniq
separately ^_^), but I'm not sure that this is really what we want. Each individual pipe link is something that we might want to change by editing some page.
Incidentally, I don't think that anybody ever wrote categorys
; that's always been categories
. (On Wikipedia, I would sometimes write categories
, categorial
, etc and create a redirect at categori, but that's borderline for MediaWiki and not wise here, I think.)
Hi Zoran,
Thanks for your comments. Having redirects definitely helps with a lot of issues, but trying to retroactively fix things that would have otherwise naturally evolved if redirects had been around since the beginning is a painful process. For example, when you look at a page and see a bunch of links that look like pages, you might be inclined to continue writing links to additional pages to keep things "homogeneous". But the reason things originally appeared as pages was not out of preference, but due to us not having redirects in the beginning. I'm hoping that if I change enough links to pages, then when you look around, the "homogeneous" thing would be to write additional pages. We're not there yet, but getting there.
I'm REALLY happy to see you start to take advantage of redirects. If you add content while taking advantage of redirects from the beginning there will be no reason for me or anyone else to follow behind you making minor changes, creating additional revisions, etc. You're feedback is very valuable. Thanks!
Now I just need to get Urs on board. I still spend a lot of time following behind him. But I'd rather, and this applies to you as well, see people add content freely without worrying too much about formatting and let others (if they care to) follow behind and clean things up.
On the topic of this thread, I actually do continue to think we need a "Matters of Style" page and I would argue that we already do have a de facto MoS that isn't written down yet. Toby, for example, and me to a certain extent are already acting as "moderators of style". I admit that I am frustrated seeing a format change I make changed BACK. From now on, if anyone changes a formatting change that I make, I would like that person to point to the relevant section of the MoS so that I will know WHY my formatting was undone.
I suggest that any format change (and that goes obviously for me as well) that is made should be justified by referring to a relevant section of the MoS. My time is valuable and to see my format changes undone makes me feel as though I wasted time. To minimize the chance that I waste my time, I'd like to see some rules.
Like I said, there are rules already (or else my format changes would not be undone!) so we should write them down.
I've undone a few of Eric's style edits on the basis of these three rules:
Relative to a MoS, these are more meta-rules than rules. The naming conventions have rules, and these are in a sense ‘incorporated by reference’ in (2), but the place to write those down is in a page on naming conventions. And my meta-rule (1) for reversion is to enforce the idea that there are no (non-meta) rules to enforce! (The other one was also meta, I think; I'll try to remember what it was.)
So basically, my attitude is that there is no Manual of Style (although there are Naming Conventions). Maybe that should be written down somewhere!
If there are meta-rules, these meta-rules should be enunciated so that everyone knows about them and not just a select few. The MoS does not need to be set in stone (as I've said), but should represent a set of guidelines to help avoid unintentional gendarmes. You obviously have some meta-rules in mind and I'd like to know about them so I do not waste my time making changes that will only be undone. That is what I mean by an MoS.
foos
to foos
, so if you change foos
to foos
and nothing else, then I might change it back, particularly if I know that Zoran wrote it. (Although you are probably not changing it anymore if you think that Zoran wrote it.) Do you want to record on a MoS page those stylistic differences that some people prefer so that everybody knows that they're not mistakes?I just told you my meta-rules, which are simply my personal rules for reverting those few edits of yours that I've reverted. It's fine with me if you want to record them somewhere. But there's nothing in them that states that people ought to write things in some particular way, which is what a Manual of Style means to me. (Except for the naming conventions, which are written down already.)
No worries. It sounded like you had more in mind than "]]s" vs "s]]" wrt meta-rules. I thought there was more than that.
Oh! I remember. On additive functor, it was either you or Mike that had a link Cauchy colimits. Since I came across several pages with wanted links for Cauchy colimit AND since someone already thought Cauchy complete category was a good place to land after clicking Cauchy colimit, I added redirects
!redirects Cauchy colimit !redirects Cauchy colimits
to Cauchy complete category. I then went around and did my usual plural Cauchy colimits to Cauchy colimits cleanup routine. Then you (I think) deleted my redirects. This broke all my plural links.
I even used that as the example on redirects hoping someone would remove the redirects from Cauchy complete category and create a page Cauchy colimit. However, you went ahead and removed the redirects (breaking all my plural links), but didn't create Cauchy colimit. Then you deleted any reference to this example from redirect in place of your own example.
The end result is we have broken plural links for Cauchy colimits with no evidence of the work I did. Hopefully, someone creates that page and we can add the plural redirects to fix the links (rather than changing to Caucy colimits).
It's not a huge deal, but when work is undone, I wish there were some explanation given to help avoid wasted effort in the future.
PS: I really don't mind my work being undone if it improves the quality of the nLab, but my only request is that some remark is left somewhere, preferably an MoS giving the reason.
The end result is we have broken plural links for Cauchy colimits
Where are there links to Cauchy colimits? I checked and didn't find any; I just checked again and didn't find any. I did leave some wanted links to Cauchy colimit, which I think is what they should be.
In any case, my reversal of that particular redirect was not for any style reason. I explained it at latest changes; you must have missed that. (Incidentally, Google has no hits for "Cauchy colimit" except on the Lab. Perhaps the original writer meant to say ‘absolute colimit’? It might be reasonable to redirect that to Cauchy complete category, at least temporarily.)
If you want a meta-style rule out of this, then it would be that just because somebody writes bar
somewhere doesn't mean that !redirects bar
belongs at foo; one must read foo to see if that's appropriate. However, I don't think that I would call that a matter of style at all. Still, perhaps that was what I was planning to put as (3), since I did revert you on that in this case.
(PS: The redirect cache bug still exists and may be seen at zero object. I suggest that nobody edit that page until Jacques has a chance to verify it.)
No worries. That again is no biggie.
In my defense, it was not bar, it was Cauchy bar :)
(Jacques has now seen zero object. It turns out that the cache bug upon removing a redirect would be very difficult to fix. Fortunately one can clear it by hand by editing the page.)
I see a general issue becoming relevant, independent of whether we are talking style or content or otherwise:
namely, in general it may happen that new changes made to old material might not appear entirely as an improvement to the former contributors, one way or other.
I can't see how we can find a general rule to deal with such issues. I think we have to fight it out -- of course: discuss it -- on a case by case basis.
Everybody must know that nothing he or she enters into the nLab is guaranteed to remain there permanently unchanged.
On the other hand, everybody should get the impression that everybody else is trying their best to make changes only where they actually lead to an improvement of sorts.
This will be a group process. I wouldn't know how to organize or formalize it. If we run into serious conflicts on how an entry should look like, we need to open a query box there and have a discussion until some kind of compromise is reached.
I agree with all that Urs has said.
Perhaps I should have opened a query box at Cauchy complete category insted of (or in addition to) explaining on latest changes? Yes, I think that I should have!
Sounds good to me too.
Sometimes I will leave a note in a query box indicating that I will remove the query box soon. I'm not sure if that is the right thing to do, but I think if an edit is there for 20 minutes or so, it will show up in the history. Once I'm sure the note will appear in the history, I remove the note. That way, if someone traces back through revisions, they will see the note.
I think it is a good idea to leave little query notes explaining a change, especially a format change, so the author will have the option to disagree or to agree and possibly change their style.
I have an idea. I might create a page Note on formatting (or something) with a few comments about formatting changes I/we generally make. Anyone is welcome to modify it. When I make a format change to a page, I will provide a link to this page. The link will be there only temporarily. The page itself will keep track of pages pointing to it and I will periodically go back around and delete the links to the note. This will produce a few spurious revisions, but in the scheme of things, I don't think that is a big deal. This could also be a place to explain specific examples of format changes that are made. In this way, when an author sees that his/her content has been changed and sees a link to Note on formatting, in principle, they can go there to see why the format was changed and can either agree and leave the change or disagree and undo the change, but at least they will know "why".
The Note on formatting is not a set of rules, but a set of explanations.
I like the idea of explanations for potentially minor formatting issues. It's much better than a guidelines page for something that's so much a matter of taste.
However, there is a slight irony in the wording, given who wrote it! All along it refers to a mysterious "we" who go round (presumably at night) cleaning up the n-lab. Does the n-lab now have "house elves"? (Or should these be "lab elves"?).
Two suggestions. Somewhere prominent, there should be a note saying that if you see a minor formatting change of something you wrote then you should scroll back through the history to see if someone left a comment on why it was done, otherwise people new to wikis might not be in the habit of going back through revisions and so would never spot the note or link. I'd put this in the FAQ as it seems like the most likely place someone would go to find out about such matters.
Secondly, how about an obvious icon as the link rather than the page name. That would make it visually clear that this is something a bit different to an ordinary link on the page. Under the influence of "Eats, shoots and leaves", I would go for a big "," [comma].
Of course we have lab elves :)
I like the idea of a special character rather than a page name for the links to Notes on Formatting. If it is unicode, we can create a redirect for it.
We need a (unicode) symbol for the lab elves :)
I like the wheel, not the knight.
Yes, I like explanations much better than rules!
I think if an edit is there for 20 minutes or so, it will show up in the history
It's 30 minutes.
I had in mind something a little more eye-catching. Something maybe 3 or 4 times the size of a standard character. As well as being more obvious, it's size would make it clear that it's something "meta" and not intended as part of the page. I think that there are other situations where such icons are warranted, marking "original research" is another.
Yes, we should mark original research, in order to warn the reader but mainly in order to encourage the reader -- and us -- to participate in research activity on the nLab in an organized way; meaning: in a way that doesn't interfere but cross-fertilizes with the expository parts of the Lab.
At the entry twisted cohomology I added a standout box to indicate original research.
I think we should have more of that kind.
1 to 39 of 39