0

I noticed today that tags can't contain characters with diacritics symbols (I tried to add the "túró" tag and I got the "tr" tag: the software just removed the unallowed characters without notice) so apparently we can't use the correct spellings when non basic Latin characters are used (i.e. crêpe, brûlée, spätzle, főzelék, lecsó, döner, frappé, Þorramatur, Kåldolmar).

Is that true? Is there a reason why only ASCII characters are allowed in tags?

By the way, I would never eat a dish in a restaurant if its name is misspelled - if the name is wrong, what about its ingredients or preparation? Cooking requires care about details.

  • I think all of those tags seems too specific to be included as tag in their own right, and would all fall under some other more general tag, like cheese dessert etc. Admittedly I'm not familiar with all the terms so maybe there would be an argument to be made... – Sam Holder Jul 27 '10 at 8:35
  • 1
    I think that a "crêpe" tag make sense (there are many crêpe variants, both dessert and main dishes). Now is in use the "crepe" version without the diacritic on the central "e". – Wizard79 Jul 27 '10 at 9:26
  • yeah, maybe that is true Lorenzo. I suppose not having a current example is still no reason to not want to allow this. – Sam Holder Jul 27 '10 at 9:28
  • I've expanded my answer with my thoughts re: crepe/crêpe – Rob Jul 27 '10 at 9:36
  • Do we have any information on how the tag synonyms are implemented? There was mention that there's a process that goes through and updates tags to preferred terms, but are they also used for query expansion (if I search under a non-preferred term, will it find the items under the preferred term?) – Joe Jul 29 '10 at 4:04
  • @Joe: I just tested on Stack Overflow and it would seem that they are not used for query expansion. To me that's a bug that should be fixed ASAP, I'm going to mention it on MSO. – Aaronut Jul 30 '10 at 14:06
  • @Joe: Actually, looks like it's already been requested. I hope that gets fixed soon, as it stunt the usefulness of the diacritic-synonym MSO proposal. – Aaronut Jul 30 '10 at 14:47
2

We now remap (most) international characters to their ASCII equivalents in tags.

So instead of entering

túró

and getting

tr

you will get

turo

  • Awesome. Thanks Jeff! – hobodave Jul 31 '10 at 8:47
  • Thank you Jeff, this is fine. I would like much more if diacritics were allowed, but the remapping is an acceptable compromise and certainly a good fix for the character dropping issue. – Wizard79 Jul 31 '10 at 12:18
1

Update:

I've put in an official feature request with the Stack Exchange team. I think that as long as we do this right, making effective use of the Tag Synonym feature, most of the usability issues should simply vanish.


P.S. I considered keeping the original text of this answer around for continuity, but the opportunities afforded by synonyms make it essentially irrelevant now. Bring on the diacritics with automatic synonyms!

  • 1
    You bring up a good point that tags are used for searching, but at the same time, string comparisons in many programming APIs can be "culture insensitive", so that a search for crepe can still yield crêpe. Thus, if programmed appropriately, you shouldn't need an int'l keyboard or a character map to use diacritics well within tag names. Granted, it's still an issue when creating a new tag, but that's a one-time fix if done incorrectly the first time. – Ben McCormack Jul 27 '10 at 16:05
  • 1
    @Ben: Even if the search takes culture info into account (not always as easy as it sounds), the issue of creating new tags is actually not a one-time issue, since North American members will continually and repeatedly type the tag without diacritics, ignoring the autocomplete. It'll be a never-ending retagging and tag-merging fiasco. We're really better off with the lowest common denominator. Everybody can understand "no diacritics/accents" - only some people can understand the converse. – Aaronut Jul 27 '10 at 18:28
  • 1
    @Aaron fair enough. In the grand scheme of things, I would consider it pretty low on the priority list. However, I think properly-implemented (properly-implemented being the operative phrase) diacritics would be a fantastic tiny little detail to add to the site. – Ben McCormack Jul 27 '10 at 18:42
  • I'm glad a Canadian said this. :D I think if I had said it, I would come across as an arrogant American. – hobodave Jul 27 '10 at 22:02
  • 1
    @Ben - I agree entirely with the properly-implemented*/*fantastictiny little detail part of your comment, if a community caters to the lowest common denominator then it's never going to achieve all it can. If everywhere on t'internet starts using "crepe" instead of "crêpe" for example, pretty soon the only places where it'll be spelt correctly will be french speaking countries. That'd be a very sorry state to be in. – Rob Jul 28 '10 at 8:55
  • @Aarnought: so ya prefer wehn piple missspels uords? Don't forget that the same word may mean two different things with different diacritics. This is the classic arrogant position on localization of English-centric-people. But remember that the world goes well beyond ASCII. – Wizard79 Jul 28 '10 at 13:33
  • 1
    @Rob: That is ridiculous. We are only talking about the tags, not the titles or the content. Do you really not see the difference? – Aaronut Jul 28 '10 at 14:45
  • 3
    @Lorenzo: See my comment to Rob. This has nothing to do with encodings or arrogance and everything to do with usability. The limitation is only in the tags; as you've obviously noticed for yourself, diacritics are fully supported everywhere else. It's not even just English; not every country uses the same diacritics. Can you type Cyrillic characters on your keyboard in case we start getting questions about Russian food? – Aaronut Jul 28 '10 at 14:47
  • 3
    And @Lorenzo, I think you fall into the same category that hobodave was criticizing here. Your tone seems to indicate a concerted effort, conscious or unconscious, to assert superiority over those mouth-breathing Americans, i.e. by referring to the "arrogant position" of "English-centric people" and "misspelled" words. But you've failed to address the glaringly obvious usability issue and ignore the fact that, statistically, there are far fewer people who can easily type diacritics than those who can't. – Aaronut Jul 28 '10 at 14:54
  • @Aaronut, there's really no need to be patronising. Of course I can distinguish the difference between tags, titles and content. My opinion, which is as I said, my opinion, is that it would be better to accept tags that are correct, rather than their anglicised equivalents. – Rob Jul 28 '10 at 17:55
  • @Aaronut: Yea I find @Lorenzo's tone rather offensive. @Lorenzo: You've alluded to the fact that you feel disadvantaged before here: meta.cooking.stackexchange.com/questions/236/… Your attitude and statements are coming across as if you want a little section of this site to be catered specifically to you and the people of Italy. I'm not sure if the topic of what languages you use has come up on a SE before, but if you feel that disadvantaged over the perceived disrespect we have for your culture, perhaps you should start ... – hobodave Jul 28 '10 at 17:57
  • @Lorenzo: ... a SE specifically for Italian Food & Cooking. – hobodave Jul 28 '10 at 17:57
  • That's an interesting thread @hobodave, I hadn't really read it in detail before. This may be the first Stack Exchange site where the effect of cultural and geographical divides may actually be very significant, and I can see the need to be more mindful of localized assumptions (which, unfortunately, most of us don't even know we're making). That said, there's a major distinction between cultural/geographical and language issues. English is the lingua franca of the web, and it would be disingenuous to claim that "anglicization" of tags presents a serious problem. – Aaronut Jul 28 '10 at 19:10
  • @Aaronut: I think tag synonyms is the way to go, assuming they can get support for more characters and searching built-in. I suspect requiring moderators to enter the tags with diacritics and set up synonyms is the most workable solution. – Shog9 Jul 28 '10 at 19:24
  • @Knives: Well, we can't stop users from creating the tags themselves, with or without diacritics. So people would have to be vigilant about editing the tags to be "correct" and then creating the synonyms. If moderators can do this without votes, that's great; but I'm wary of depending on mods (especially since it's not clear when we'll have them!) I definitely think there should be a rule that if you edit the first occurrence of an anglicized tag to the "proper" version, you should be responsible for proposing the synonym or contacting a moderator (through the flag or meta). – Aaronut Jul 28 '10 at 19:33
0

I suspect that the reason only ASCII characters are allowed lies in the origins of the software that powers this site. It was originally written (as I guess you know from having looked at your profile =) for stackoverflow, a programmers Q&A site, where tags are less likely to conain characters other than standard ASCII characters. So, it was probably a case of "allow only what's needed".

It'd defintely be valuable, as I can think of a few reasonable tags off the top of my head, for us to request tags with diacritics in be permitted. +1 from me.

Just following on from a point made in a comment on the original question, if we do request that tags allow diacritics, we should probably suggest that when someone enters a tag that matches exactly, bar diacritics, they're pointed towards that tag. So, if someone tags a question "crepe", they should be given the suggestion:

Are you sure you didn't mean 'crêpe'?

Otherwise we risk having two tags for crepe/crêpe and having to clean them up.

  • At least, they should not allow to submit tags with unallowed characters, because now the software is just removing them, resulting in nonsense tags (like the "tr" I got for my "túró", that I then edited to "turo"). – Wizard79 Jul 27 '10 at 9:29
  • 5
    Jeez, you'd think the creators of this QA engine would have read the article, "The Absolute Minimum Every Software Developer Absolutely, Positively Must Know About Unicode and Character Sets (No Excuses!)" joelonsoftware.com/articles/Unicode.html – Ben McCormack Jul 27 '10 at 12:39

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .