5 ms·
It's 2016 and Bootstrap still doesn't have native RTL support. Foundation had it (excellently) since 2013. When I'm building something for the local market, thi
by yoavm 11y ago
It's 2016 and Bootstrap still doesn't have native RTL support. Foundation had it (excellently) since 2013.
When I'm building something for the local market, this has a major affect on my decision.
- ZeroGravitas 11y agoThere was an attempt to do so. They were using an RTL-flipping tool and it all looked pretty cool (I built something with it) but then one of the people who manually flipped Bootstrap basically accused them of being culturally insensitive by trying to automate this task and they abandoned it.
- reitanqild 11y agoIf true this could be added to a list of nasty side effects of rampant enforcement of political correctness.
- frogtog 11y agoIt's not true. The user didn't accuse them of being culturally insensitive. Rather, he argued that that automating the conversion was fool's errand, partly due to the differences between the various RTL languages. He suggested that the Bootstrap team point users to quality RTL themes instead of releasing a shoddy built-in RTL implementation. Perhaps this should be added to the list of instances of people dismissing valid arguments by labeling them polical correctness.
- reitanqild 11y agoOk, thanks for the clarification :-)
- etimberg 11y agoI think this [1] is the related thread. 1. https://github.com/twbs/bootstrap/issues/13564 https://github.com/twbs/bootstrap/issues/13564
- harlanlewis 11y ago"Culturally insensitive" wasn't part of the discussion. He listed specific technical and UX issues with their approach. Another commenter in this thread posted a link to the github discussion.
- ZeroGravitas 11y agoHaving re-read the thread, I stand by my "in-a-nutshell" summary of it. I don't think any of his technical arguments hold up. I'm also not sure that he would have convinced the Bootstrap team to have dropped this based on technical arguments if he hadn't seemed so annoyed, and made repeated references to how only people who speak the language should make certain decisions. Some examples: * the bug report is a command: "Stop developing RTL for Bootstrap!" * the bug report ends with an exclamation point! * the bug starts "I'm really annoyed..." * "use cases that only natives can understand" * A second command, bolded: "Remove it!" I personally was inconvenienced by this because I was writing lots of bootstrap based .less, and if twitter and Bootstrap had worked together on an auto-flipping rtl system for Bootstrap then all the RTL users of my code would have got a far better service. Instead only those who use the Bootstrap CSS essentially as-it-comes get a manually RTL'd version from a 3rd party. A tool that can automate that work sounds like a far better deal for the RTL world, even if it would have annoyed this particular person. I note that his argument against automated flipping included the fact that you really need one for farsi, one for arabic, etc. Yet he only manually produces one. Maybe with the right tooling he'd have got to where he wanted to be faster. A link to such a tool: https://github.com/MohammadYounes/rtlcss https://github.com/MohammadYounes/rtlcss (note how it documents various solutions to the 'problems' listed in the original bug report, like changing the font stack when in RTL mode)
- harlanlewis 11y agoAgreed that it wasn't raised very diplomatically. That's part personality/choice, part what happens when English isn't someone's first language. Comes with the localization territory. That he produced a coded alternative, regardless of its merits, puts him in good faith in my book. There's no shortage of insufferably squeaky wheels who aren't acquiesced to, and I don't see any ways this one was treated exceptionally. It looks like the maintainers worked to understand the root of his frustration despite his abrasive style. Kudos. The Bootstrap team acknowledging it's a complex issue (rather than the solved problem they thought it to be) and changing their release plan accordingly seems a reasonable way to handle it, and one I'd encourage for maintainers of widely used projects. Finding the right effort/quality balance is tricky. That some of the issues could be relatively trivial to address (font stack) doesn't invalidate the category. In particular, the distinction between RTL and “flipping” does seem to be a meaningful one, and fraught with unexpected issues. Eg, next=right and prev=left as a cultural convention that doesn't directly translate. As you note. Frankly, if tasked with supporting RTL I'd probably start with a repo like the one you linked to, but fully expect it to fall down in a number of ways that could only be identified by a native user. Whether those issues get addressed would be a pretty clear statement on how much those users are valued. PS upvoted for pleasant discussion :)
- ZeroGravitas 11y agoApparently the current status is to do this in a minor update just after 4.0 ships: https://github.com/twbs/bootstrap/issues/19555#issuecomment-197549158 https://github.com/twbs/bootstrap/issues/19555#issuecomment-...