Bugs, Requests, and Changes
Anthologica Universe Atlas / Academia / Site Help and Bug Reports / Bugs, Requests, and Changes

? twabs fair maiden
posts: 211
, Conversational Speaker, [ˈaɪwə]
message
also, evidently, it does not handle quote blocks correctly if there is a ] in the quote source?

quoting metaphysical questions:
what is god

[quote [metaphysical questions]]husserl doesn't care
? Rhetorica Sleepless Scribe
posts: 1204
, Kelatetía: Dis, Major Belt 1
message
1. That's yo' browser, foo'.
2. Regexes are not context-free grammars. It would be very impractical to implement a full CFG-based parser for tags on Annie, and very slow.
? twabs fair maiden
posts: 211
, Conversational Speaker, [ˈaɪwə]
message
How easy would it be to simply allow the tag to take an argument in quotes instead? Shouldn't be too hard, but it's also not at all important.

Honestly, if you provided me with the Cadre source, I'd even be willing to write it myself.
? twabs fair maiden
posts: 211
, Conversational Speaker, [ˈaɪwə]
message
quoting Rhetorica, Kelatetía: Dis, Major Belt 1:
User profiles will now refuse to let new blog posts be made if one called 'Untitled Entry' already exists. In unrelated news, salamanders are incredibly lazy creatures.

Up until now I was thinking "why is this my fault; also your 'solution' is the laziest thing I've seen", then I realized what Annie is, viz: a wiki that simply doesn't look like one.

Which makes this behaviour make a whole lot more sense.

That said, it still doesn't make as much sense as it should. For example: Why do the blog posts act differently then every other kind of post (with the... partial exception of forum posts?) Instead of prompting you for a page name and then giving you the fields for the page content, it creates an empty post with the title "Untitled Post" and then gives you a screen for editing that post. This is evidenced by the fact that if you close that page without saving, the empty "Untitled Post" will still exist.

Other kinds of articles have similar behaviour, viz. you create an article and are then brought to a screen for editing it. Which makes sense, from some perspectives, but it has the unfortunate consequences of (1) you have to have a separate screen just to give your article a title, (2) if you decide after that first screen that you don't want it after all, you have to explicitly go and delete the article you just created.

On the other hand, I'd expect given this model that the URL I get sent to after the first screen be e.g. http://anthologi.ca/?id=257839&mode=edit. But it's not, it's just http://anthologi.ca. Is there rerouting involved here?

One odd behaviour which I don't think was intentional is, if you change a page title and hit 'save', the SUCCESS screen has the *old* page title and not the new one.

Another odd behaviour is, if you rename an article to the same title as an existing article, then it won't be renamed, but all of the content will be lost.

The wiki model makes sense in many respects, like, there's no point creating two articles with the same title, so we can do that here. But it does fail in others, especially if there is only one namespace (and I'm presuming there is, because you don't give one in the URI—but then again the function which determines what page to edit must be checking namespaces somehow, probably just in the form of authors, because I can create a blog post called "Big..." with no problem.) When we get to the matter of dictionary entries, is necessarily, well, necessary to be able to create two things with the same title, and the way in which the site handles this—wouldn't be unnecessarily klugy on its *own*, but we already *have* a function for resolving namespace disputes when the author isn't implied, so why don't we use it?

I'm not sure if hitting "reply" to a forum post is creating an article in the same way—if I close the page it doesn't seem to be created, and I can't locate it through the search function either, so probably not, but if that's the case, well, I don't understand, why don't we do this with other things?

I'm sorry that I'm being critical like this; I know that Cadre largely works and everything that I've complained about, both in this post and in the past, has been stupid things that I should just be able to work around. And I don't want to naysay inelegant solutions—I'd be all kinds of hypocritical if I did. But if the things I've been talking about aren't just, the way they are for a good reason, then I'd honestly be willing to do the work to make them better. If only for my own benefit.
? Rhetorica Sleepless Scribe
posts: 1204
, Kelatetía: Dis, Major Belt 1
message
Right, you're really prying up some of the wallpaper here. Let's see what I can do about some of them:

a wiki that simply doesn't look like one.

Correct. Almost correct. Cadre is a recursive wiki; every page must have a unique parent/title combination. Like a filesystem, you can't have two files with the same name in one directory. (And everything is potentially a directory.) Forum posts use a different workaround; every single one has its timestamp encoded as part of its title to ensure it's unique. Your post is titled Bugs, Requests, and Changes (2016-10-03 17:33:57) , and if there were a conflict, a random number would be appended by the Annie-specific forum code to address this.

especially if there is only one namespace (and I'm presuming there is, because you don't give one in the URI

There actually is a way of referring to page titles with context, but no code does this and it's ugly internally. http://anthologi.ca/users/rhetorica will look for a page called 'users' with a child called 'rhetorica', which allows semi-unique title referencing for pages that might otherwise be masked by lower-IDed ones with the same title. This is still handled by index.star, as you'll note you get redirected to a page with a question mark in the URL.

When we get to the matter of dictionary entries, is necessarily, well, necessary to be able to create two things with the same title, and the way in which the site handles this...

This is gracefully handled by the use of page parents to delimit namespaces. If you absolutely need to have multiple entries with the same name in one dictionary, there's some code to automatically add <!>1, <!>2, <!>3, etc. after the name, which is visually represented as (1), (2), (3), etc.

Instead of prompting you for a page name and then giving you the fields for the page content, it creates an empty post with the title "Untitled Post" and then gives you a screen for editing that post.

This is an interface blemish and can be fixed. "Post without title" was the habitual design choice in very early (2009-2010) Cadre templates because I thought it was really slick to be able to embed the edit mode of the child back when I didn't have an easy shortcut for adding page redirects. This code is definitely overdue for replacement, as it creates unexpected and undesirable outcomes during interface exploration. (Many Annie blogs have ended up with blank Untitled Entry posts over time.) There are a number of solutions to this problem, such as manual title entry prior to prompting (typical of templates based on the venerable Abstract List, such as the Dictionary) or duplicating the edit mode code into the parent template (which is the solution used for blog comments and forum posts.) In general, inconsistency like this in an interface is evidence that true code reuse is scarce.

On the other hand, I'd expect given this model that the URL I get sent to after the first screen be e.g. http://anthologi.ca/?id=257839&mode=edit. But it's not, it's just http://anthologi.ca. Is there rerouting involved here?

This is an HTTP thing. Because the page is submitted with what's called the POST method (rather than the GET method), all of the parameters (the stuff after the ?) are passed using a different field instead of the URL field. Both the URLs http://anthologi.ca/ and http://anthologica/?id=257839&mode=edit are handled by one script, http://anthologi.ca/index.star.

One odd behaviour which I don't think was intentional is, if you change a page title and hit 'save', the SUCCESS screen has the *old* page title and not the new one.

This is a bit of a dumb detail; the page's title is cached for display prior to being updated. There are several different ways for Cadre to change the metadata of a page, and few of them update the $pagetitle variable actually used by the template. I'm pretty sure I could fix this.

Another odd behaviour is, if you rename an article to the same title as an existing article, then it won't be renamed, but all of the content will be lost.

I'm not certain what the deal here is; that's a genuine bug report. It probably has something to do with extended object data (containing the source page's revision history) getting transferred to the target page.

I'd honestly be willing to do the work to make them better. If only for my own benefit.

...No one who says this is ever telling the truth.

But, start here: http://code.rhetori.ca/Writing_Templates
? Hallow XIII Primordial Crab
posts: 482
, 巴塞尔之子
message
the spirit is willing, but the flesh is weak
? twabs fair maiden
posts: 211
, Conversational Speaker, [ˈaɪwə]
message
100% true, although I meant that in a different context when I last said it1

But on the other hand, I have done that kind of thing (programming for someone else's sake, or something) before—indeed right now I am rewriting an Ancient Greek transliteration module, just because it is badly written.

_________________________
1. and maybe you are alluding to that, or maybe not. It was a comment on how I can go to speak or sing publicly without (I think) feeling any fear, but my knees shake and I sweat. Which fact is very interesting to me.
? Rhetorica Sleepless Scribe
posts: 1204
, Kelatetía: Dis, Major Belt 1
message
The Updates section on the front page side bar should no longer get stuck displaying old words or articles due to the priority flag being set on those pages. (The priority flag is used to make things 'sticky' in lists, like forum topics, and can be set on articles with the 'highlight' checkbox. I have no idea why words have been getting this set on them.)
? Rhetorica Sleepless Scribe
posts: 1204
, Kelatetía: Dis, Major Belt 1
message
Search finally fixed. Traced the problem to operator precedence in the page type filtering. Added parentheses. Rejoice!
? Hallow XIII Primordial Crab
posts: 482
, 巴塞尔之子
message
*applause*
? Izambri Left of the middle
posts: 740
, Duke, the Findible League
message
*throws a party*
? Serafín posts: 36
, 農, Canada
message
Not sure if it's been reported before, but I can't export a dictionary in print-friendly format.

"Fatal error: Allowed memory size of 134217728 bytes exhausted (tried to allocate 130968 bytes) in /home/web/cadre/libraries/htmlpurifier/HTMLPurifier.standalone.php on line 7720"
? Rhetorica Sleepless Scribe
posts: 1204
, Kelatetía: Dis, Major Belt 1
message
Well done! Your ambition is probably excessive. I'll look into it, and/or ways of selectively disabling the HTMLPurifier extension for that mode.
? Rhetorica Sleepless Scribe
posts: 1204
, Kelatetía: Dis, Major Belt 1
message
? twabs fair maiden
posts: 211
, Conversational Speaker, [ˈaɪwə]
message
I think you said you fixed searching but it's still not quite fixed, seems to need the author_restrict=1&group_restrict=1 parameters to work right.

also, if you care, the ellipses in the search results are added unconditionally
? Rhetorica Sleepless Scribe
posts: 1204
, Kelatetía: Dis, Major Belt 1
message
This query works just fine without those parameters. Can you give an example that breaks? (Fixed the extra ellipses, though.)
? twabs fair maiden
posts: 211
, Conversational Speaker, [ˈaɪwə]
message
If I search for forum posts owned by me containing, say, "one", I get

http://anthologi.ca/?id=48&q=one&go=Search&author_restrict=1&author=15&group=0&type_restrict=1&template=41

which returns not only forum posts but also dictionary entries and maybe other things.

More pressingly, the "ex foris" link gives me everything on the forum:

http://anthologi.ca/?id=48&q=&go=Search&name_restrict=1&author=15&type_restrict=1&template=41

Adding author_restrict=1 gives every forum post and dictionary entry and whatever else I made. (This is kind of odd since without the author_restrict parameter it seems to just give forum posts.) Adding group_restrict=1 fixes this as well.

Also, the bug where one of the page links is replaced with a post title is still there.
? Rhetorica Sleepless Scribe
posts: 1204
, Kelatetía: Dis, Major Belt 1
message
Done and done! name_restrict should be author_restrict; I changed the name of the parameter but forgot to fix the user profile template to reflect this. Secondly, the 'author + type' restriction case was implemented to act like the 'author' restriction case; a quick variable name substitution solved that. Hopefull search will genuinely not suck now.
? Serafín posts: 36
, 農, Canada
message
In Diccionario alaia-español, I can't see any tags at the bottom to click on, even though when you look at the data of any word you can see that the word class tag is there.

http://anthologi.ca/?id=259344
? Izambri Left of the middle
posts: 740
, Duke, the Findible League
message
Perhaps not the best place to ask it, but...

Why do we have that here?