Search result for infix: / is a 404

[coke]

It points at http://doc.perl6.org/routine/%2F

[zoffixznet]

Seems there's a whole bunch of them that show up in the results, but result in a 404 when clicked. E.g. .+ (http://doc.perl6.org/routine/.%2B), .? (http://doc.perl6.org/routine/.%3F), .* (http://doc.perl6.org/routine/.*)

[zoffixznet]

The / and // here too: http://doc.perl6.org/routine.html These might require enabling Apache's EncodeSlashes or whatever that option is called.

http://irclog.perlgeek.de/perl6/2015-12-01#i_11635192
http://irclog.perlgeek.de/perl6/2015-12-01#i_11635196

[autarch]

It'd take more than just EncodeSlashes. Some of these may be blocked because of starting with '.'. It might be better to have htmlify.p6 rewrite all symbols into names like 'slash' and 'dot-star', etc.

[zoffixznet]

OK, part of the problem is Apache by default is configured to ignore dotfiles, so that's why all the .whatever operators are 404s.

Since changing IndexIgnore directive alone doesn't fix the entire issue, I think we should seek some alternate solution anyway. Since we can't possibly have a file named '/.html', maybe we should do something like have a base64 encoded filenames that are mapped to from URLs by mod_rewrite + a short decoder script.... just throwing ideas out.

[pdl]

I had a go at trying to work through these locally last month and gave up. There are a lot of these.

• / is bad because it's the path separator
• . is bad because of dotfiles, also .. for parent directory; and who knows what else might be going on with file extensions
• % is bad because of percent-encoding (but works)
• ? indicates parameters begin
• & separates parameters
• = separates keys and values in parameters
• # indicates the request part is over, the rest is the anchor
• You also want to be careful with + because it gets percent-encoded; if that ends up in parameters, it may be treated as a space.
• \ ...actually seems fine

It would be really great to be thinking about SEO at this point, because punctuation-only variables/operators/whatever are a) hard or impossible to google; b) prone to shell-escaping error (perldoc -v $_ on bash) - and @autarch's suggestion would certainly help in this department.

[zoffixznet]

Oh, I seem to have missed @autarch's suggestion to convert symbols to words. Yes, a definite 👍 on that.

[AlexDaniel]

Yep, way to go.

[AlexDaniel]

It might be better to have htmlify.p6 rewrite all symbols into names like 'slash' and 'dot-star', etc.

We can borrow these names from this list: http://www.muppetlabs.com/~breadbox/intercal-man/tonsila.html

[stmuk]

Or just "url encode" then into the % form,

On 28 January 2016 at 04:28, Aleks-Daniel Jakimenko-Aleksejev <
notifications@github.com> wrote:

It might be better to have htmlify.p6 rewrite all symbols into names like
'slash' and 'dot-star', etc.

We can borrow these names from this list:
http://www.muppetlabs.com/~breadbox/intercal-man/tonsila.html

—
Reply to this email directly or view it on GitHub
#155 (comment).

4096R/EA75174B Steve Mynott steve.mynott@gmail.com

[zoffixznet]

I rather use names, as they're more descriptive and, on a wild guess, would be picked up by eearch engines better. Things like "." still remain "." in percent encoding, so you'd have to add custom rule anyway.

[AlexDaniel]

I think that @jsoref is making some progress regarding this issue. Contact him if you want to help.

[gfldex]

here is my list.

$ --> $dollar
/ --> $slash
\ --> $backslash
. --> $dot
% --> $percent
? --> $questionmark
& --> $ampersant
= --> $equal
# --> $hash
+ --> $plus
" --> $quote
' --> $singlequote

We might even use ´'$' ~ $.uniname.subst(' ','')´ ( there is an underscore in the 2nd param) so we only have to keep a list of chars we want to escape and the actual names are generated by Perl 6 for us (what saves time to find typos). I will extend Pod::To::HTML by a callback to filter links as a first step.

[zoffixznet]

It's worth noting teatime was working on the fix for this. I've not seen them on IRC for some time though.

[zoffixznet]

Err, I meant timeless not teatime... 😹

[gfldex]

: --> $colon # $@({ Windows!

[gfldex]

~ will not play nice on some webservers and can confuse shells.

[gfldex]

< and > are also bad on windows

[gfldex]

all problems know to me seam to be fixed in branch GT#155 https://github.com/perl6/doc/compare/GT%23155

[gfldex]

fixed with dc0fabb