## Not signed in

Want to take part in these discussions? Sign in if you have an account, or apply for one below

## Site Tag Cloud

Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.

• CommentRowNumber1.
• CommentAuthorzskoda
• CommentTimeDec 28th 2009

I protest again that Eric again reedits the plurals which I created into his style plurals. E.g. connections into connections in the entry regular differential operator.

• CommentRowNumber2.
• CommentAuthorMike Shulman
• CommentTimeDec 29th 2009

On some things I think it is good to have a consistent style. For instance, a while back we had a discussion about whether to use Unicode in page names. For now we seem to have settled on "no," and even the people who were in favor of "yes" are adhering to that convention.

Now of course there's a question about where to draw the line between having an "nLab style" and allowing authors to have personal styles. But I personally find connections much uglier than connections, and although I will occasionally type the former due to being too lazy to add a missing redirect, I am happy when someone else adds the redirect and fixes them to the latter. I recall that you said something at some point about preferring the former, but I don't remember where, and right now I can't imagine what could be better about it. What do other people think?

• CommentRowNumber3.
• CommentAuthorEric
• CommentTimeDec 29th 2009
I similarly protest to your intentional use of "s" outside brackets ;)

If I see something unattractive on the nLab, one of the few contributions I can make is to beautify it. If I see an "s" outside brackets, I agree with Mike and think it is ugly and assume was due to reasons along the lines Mike gave, so I fix it. If I happen to notice it is from one of your pages, I try to ignore it because I know you have a unique taste, but to be honest, I'm not overly concerned about that.

Whether we like it or not, aesthetics is an important part of any project like this.
• CommentRowNumber4.
• CommentAuthorTobyBartels
• CommentTimeDec 29th 2009

I though that Eric had agreed not to change Zoran's ]]ss. Perhaps that was only me. I will try to find the old discussions.

Mike, you seem to have a habit of writing links that don't exist even when a link does and then not adding the redirect. Here is the latest example; the links now work because I added redirects.

• CommentRowNumber5.
• CommentAuthorEric
• CommentTimeDec 29th 2009
I did agree not to change Zoran's "s". That is what I meant by this statement:

>If I happen to notice it is from one of your pages, I try to ignore it

I should have said "I will" ignore it (because I will), i.e. I would never willingly change Zoran's stuff on purpose. But I have to admit, I do not check very carefully who wrote the "s", I tend to just fix it.
• CommentRowNumber6.
• CommentAuthorTobyBartels
• CommentTimeDec 29th 2009

Here is the previous discussion.

• CommentRowNumber7.
• CommentAuthorzskoda
• CommentTimeDec 29th 2009

There is a REAL harm. If I wriet an entry I myself like to use it. When use it, I use it most often OFFLINE, that is from the copy of html of nlab. The redirects do not work of course without support so all these connections instead of connection-s do not find appropriate file in my directory. So I can not navigate within the offline nlab.

• CommentRowNumber8.
• CommentAuthorzskoda
• CommentTimeDec 29th 2009

But I have to admit, I do not check very carefully who wrote the "s", I tend to just fix it.

Eric, when you write your own contributions of new content and new links then there is no doubt: it is yours and you do not wonder who done it. If you play around s-s of other people then this is a repair job and repair is always trickier in a technical sense then the creation job which is trickier in scientific or some other sense (we need more effort of people in finding, citing and linking references for various subjects, I usually spend more than half time on that when writing a new entry, and would prefer to concentrate on content than links but if I already have them I think it is easier to me to include them than the people who improve the entry afterwards). Toby is extremely careful, to my opinion, even too careful and extremely knowledgeable about technical stuff from fonts to rights and conventions, and we all thank him for his patience to do this thing right.

After we had discussion few months ago on s-s I myself gradually switched to plural version to agree with others but after a while, pressed by problems with redirects (for example inconcistency in listing the site which cites a redirects to be listed at the bottom of the page and the problems in offline version as I quoted above), I switched back to the old version which cites the unique true page identifier (which is also good if I want to use software to download the cited pages from some page which I am currently interested, so the convenience of the true www identifies will always have a value to me, unless computers start being more intelligent than I am, what will happen with my age, when I get dementious).

• CommentRowNumber9.
• CommentAuthorEric
• CommentTimeDec 30th 2009
Hi Zoran. Wow! I didn't realize that redirects did not work offline. That is a serious problem.

Is there anyway to fix this? What if we changed the behavior of a redirect so that instead of processing something when someone follows a link, when someone inserts a redirect, it goes around and changes all the links?

For example, if there is a [ [foo] ] and someone adds a [ [!redirects foo] ] to [ [bar] ], could instiki go around and replace things like

[ [foo] ]

to

[ [bar|foo] ]

and

[ [foo|foos] ]

to

[ [bar|foos] ]

?

I can imagine several things that can go wrong with this (particularly if someone were malicious), but having redirects not work while offline is unfortunate.

One solution, painful though, would be to install Instiki so you can run a local server even if offline.
• CommentRowNumber10.
• CommentAuthorTobyBartels
• CommentTimeDec 30th 2009

What if we changed the behavior of a redirect so that instead of processing something when someone follows a link, when someone inserts a redirect, it goes around and changes all the links?

Nice idea, but it would definitely be bad in my opinion.

We currently have, for example, &#x5B;[!redirects sigma-algebra]] at measurable space, but it would be a good idea to have a separate page sigma-algebra, since they have other uses in probability theory. Someday I'll get around to this; in the meantime, we have the redirect. But once it appears, all links to sigma-algebra will work properly, as will links to ?-algebra once we move &#x5B;[!redirects sigma-algebra]] from measurable space to sigma-algebra. With your suggested feature, we would need to create stubs for many current redirects and might have difficulty deciding if they're needed (such as constructivism vs constructive mathematics, which are almost the same thing, although one can do the latter without believing the philosophy of the former).

• CommentRowNumber11.
• CommentAuthorMike Shulman
• CommentTimeDec 30th 2009

I don't understand what "redirects don't work offline" means. The way Instiki does redirects is that if page "foo" contains [?[!redirects bar]] then on any page that contains [?[bar]] the HTML presented to the user will be a link to ".../show/foo", not to ".../show/bar". It doesn't require server-level redirects; the redirection happens at the level of Markdown link processing. How can this not work offline?

(I'm not saying that it does work, since clearly you've tried it and it doesn't—I just don't understand why it doesn't work.)

• CommentRowNumber12.
• CommentAuthorMike Shulman
• CommentTimeDec 30th 2009

Mike, you seem to have a habit of writing links that don't exist even when a link does and then not adding the redirect

Sorry! I'll try harder not to do that.

PS. How do you put something in double-brackets here on the forum without making it a link (e.g. in a code block)? Even in code blocks my double-brackets get turned into links.

• CommentRowNumber13.
• CommentAuthorAndrew Stacey
• CommentTimeDec 30th 2009

@Mike: (That's one I keep having to look up as well). You have to use entity codes. The XML codes are &#38;#91; and &#38;#93; so typing:

&#38;#91;&#38;#91;HomePage&#38;#93;&#38;#93;


produces [[HomePage]]

I second Mike's puzzlement about the offline/online behaviour. The redirect stuff happens before the HTML pages are generated so all the links should work offline as online. Zoran, can you give an example of a page where this difference in behaviour happens? Thanks.

If we can sort this out, would that remove your (Zoran's) objection to redoing the links with the plurals inside rather than outside?

• CommentRowNumber14.
• CommentAuthorTobyBartels
• CommentTimeDec 30th 2009

@Mike: Simply

&amp;#91;[HomePage]]


will suffice if that's easier. (I always use

&amp;#x5B;[HomePage]]


which is much easier to remember, isn't it? ^_^)

• CommentRowNumber15.
• CommentAuthorzskoda
• CommentTimeJan 11th 2010

By not working offline means that if I have directory with all xhtml files it does not supprt redirects. If I had full instiki in works, it would of course work.

• CommentRowNumber16.
• CommentAuthorMike Shulman
• CommentTimeJan 12th 2010

I just did the following:

1. I ran the command suggested at HowTo in order to download an html copy of the nlab.
2. I clicked around until I found a link that uses a redirect, e.g. the link to categories on adjoint equivalence
4. It works!

Because redirects in instiki modify the generated code for an anchor link, when you download the html they continue to work. They don't require a copy of instiki running locally because they don't use any HTTP redirects. At least, that's what the theory says, and that's what I experience. Are you saying that when you download the html files using the script at HowTo, the redirects don't work?

• CommentRowNumber17.
• CommentAuthorzskoda
• CommentTimeJan 20th 2010

Exactly, I used the howto setup and the redirects did not work on 3 different computers. By the way you did not say which operative system and which browser you used.

• CommentRowNumber18.
• CommentAuthorTobyBartels
• CommentTimeJan 20th 2010
• (edited Jan 21st 2010)

I also just followed the instructions at HowTo. I didn't use a browser, I used zsh, as suggested there. (Start a terminal in any Unix-like system, enter zsh, go to the directory you want to be in, copy and paste the command). The terminal was a Gnome terminal on an Ubuntu system, but wget is pretty much the same in any Unix-like system.

This produced HTML with correct links. I looked at Mike's example in Firefox, but I also looked at the HTML by hand; it had the correct link.

• CommentRowNumber19.
• CommentAuthorUrs
• CommentTimeJan 20th 2010

It's hard to google for that. This was in some discussion on the nCafe maybe a year or so back. You should email John.

• CommentRowNumber20.
• CommentAuthorMike Shulman
• CommentTimeJan 21st 2010

Zoran, can you send or post the local HTML files that you ended up with?

• CommentRowNumber21.
• CommentAuthorAndrew Stacey
• CommentTimeJan 21st 2010

Another thought: if you run wget more than once, then it only downloads pages that have changed since it last downloaded them. To figure out which these are, it asks the server "when did this page last change?". Now it's possible that the server responds with the date that the page was last edited. But if a link on that page has changed (such as when a redirect is changed), the server might not consider that substantial enough to report and wget would not download the new page.

I don't know whether or not this is happening, but it should be easy to test.

• CommentRowNumber22.
• CommentAuthorTobyBartels
• CommentTimeJan 21st 2010

Actually, I keep getting ‘Last-modified header missing -- time-stamps turned off.’. I got that the first time, which I didn't worry about, since there were no timestamps on my end to compare things to. But now there are, yet it's still downloading everything.

• CommentRowNumber23.
• CommentAuthorAndrew Stacey
• CommentTimeJan 21st 2010
• (edited Jan 21st 2010)

Hmm, just looked at the headers sent back by the nlab and you're right: 'Last-modified' isn't there. I get:

~% wget -S http://ncatlab.org/nlab/show/HomePage
--13:06:27--  http://ncatlab.org/nlab/show/HomePage
=> 'HomePage'
Resolving ncatlab.org... 68.233.9.66
Connecting to ncatlab.org|68.233.9.66|:80... connected.
HTTP request sent, awaiting response...
HTTP/1.1 200 OK
Date: Thu, 21 Jan 2010 12:06:27 GMT
Server: Apache/2.2.9 (Ubuntu) PHP/5.2.6-2ubuntu4.5 with Suhosin-Patch Phusion_Passenger/2.2.8
X-Powered-By: Phusion Passenger (mod_rails/mod_rack) 2.2.8
X-Runtime: 27
ETag: "c54a162faa68a560e132d3e1c59dfc64"
Cache-Control: private, max-age=0, must-revalidate
Set-Cookie: instiki_session=[big long string redacted]; path=/; HttpOnly
Content-Length: 26479
Status: 200
Vary: Accept-Encoding
Keep-Alive: timeout=15, max=100
Connection: Keep-Alive
Content-Type: text/html; charset=utf-8
Length: 26,479 (26K) [text/html]

100%[====================================>] 26,479        43.57K/s

13:06:28 (43.48 KB/s) - 'HomePage' saved [26479/26479]


So that means that

1. This isn't the source of the problem with redirects not working
2. This means that the wget script downloads the whole lot each time, which isn't what is wanted.
• CommentRowNumber24.
• CommentAuthorzskoda
• CommentTimeJan 21st 2010

Toby said:

This produced HTML with correct links.

I have correct links, never complained about that. I do not have REDIRECTS working. So if a link is to true name of web page no problem but if it is alias then a problem.

I could probably investigate more nad give you more info but at the moment I am more busy with filling some content and will try to organize my software once I get some free time.

• CommentRowNumber25.
• CommentAuthorAndrew Stacey
• CommentTimeJan 21st 2010

But redirects get rendered as links so they ought to work. That's what we can't understand.

Anyway, when you get a minute to do so, email me your copy of a page that doesn't work and we'll try to get to the bottom of this.

• CommentRowNumber26.
• CommentAuthorTobyBartels
• CommentTimeJan 21st 2010
• (edited Jan 21st 2010)

If the link is to a redirected title, then it is not correct.

More explicitly, the HTML in my local copy of adjoint equivalence reads:

<a class='existingWikiWord' href='category.html'>categories</a>


The HTML in the online copy reads:

<a class='existingWikiWord' href='/nlab/show/category'>categories</a>


In both cases, the link is correct: to category, not to categories.