3 ms·
I never said I hate the 69 quotation thing. I said I hate the automatic conversion of the straight quote to the 69 quote behind my back. Imagine they automatic
by jimmies 9y ago
I never said I hate the 69 quotation thing. I said I hate the automatic conversion of the straight quote to the 69 quote behind my back.
Imagine they automatically replace the "fi" with the styled "fi" behind your back and breaks text searches and normal text files.
You can be snobby all you want, as long as you initialed it and know you don't break programs' assumptions. No one says you can't have nice shit with LaTEX and Pagemaker when you need it.
Here people who just open the file and save it, or copy the text and paste it, don't mean it, yet it happens without them ever knowing.
- LeoNatan25 9y agoRegarding fi, it just shows how broken most search engines are. As a reader, what is the difference between between fi and fi? Or between a and а? If the search engine is distinguishing those because of Unicode characters, it has failed completely.
- TeMPOraL 9y agoUnicode turns trivial string comparison problems into pretty much AI-complete ones. Same visual characters, different meaning. Different characters, but same visual meaning for the person who initiated the query. Up/down-case not well defined, and even if it is, it's not always reversible (i.e. tolower(toupper(tolower(string))) is not equal to tolower(string)). There's lots of that. The longer I live, the bigger fan of simplification and standardization I become. Like with dates and times. Timezones and DST are a fucking nightmare, because politics. And then doubly so, because people enjoy themselves with their favourite regional writing formats. At this point I'm all for enforcing ISO 8601 on every communication involving dates and times. Some people object that this is turning humans into machines, etc. So be it. Nature isn't perfect, and clear communications doesn't come to us naturally. Yet it is absolutely vital in a technological society.
- LeoNatan25 9y agoThat’s closed-mindedness bordering on the silly. ISO 8601, for instance, only represents Gregorian calendar dates and times. But what about Japanese, Hebrew and Buddhist calendars, for example? Looking at the world as only “what will my Emacs store on the drive when I paste some text into it” is ridiculous. As computers become more advanced—even mobile phones are super computers now¹—we should use that computer power to make the technology work for us, not bend over to some 1980s concepts of computing and standards. It may be more difficult for you to build such a system, I understand, but as a used, I really don’t give a damn. ¹: https://www.macrumors.com/2017/09/13/a11-bionic-chip-geekbench-scores/ https://www.macrumors.com/2017/09/13/a11-bionic-chip-geekben...
- jimmies 9y ago>I understand, but as a used, I really don’t give a damn. Yeah, Because we already changed " to 69 behind their back, let's double down on correcting the 69 so it processes like "! Let's engineer our dumb, close-minded O(n) string search and CSV parser to do some AI image recognition shit in O(2^n) to figure out when the fuck the "used" had the ` involuntarily changed to ' because MacOS or Wordpress decided it looks better. That's brilliant engineering right there. Satire aside, you realize that there is a cost of security and maintenance to over-engineer shit, it's just not just convenience, right?
- LeoNatan25 9y agoYour “used” “joke” is so … satirical. Your specific bug with CSV is a developer bug. The place where you copied the incorrect CSV format should not have had such substitution enabled. On the Mac, developer can specify for each text view and text field what substitutions are allowed by default. Likewise in web, it is possible to specify which substitutions should be allowed for text areas. So instead of blaming incompetent developers for their incorrect use of system features, thus ruining some very narrow cases, let's hold back any kind of text input and processing advancements, because you are unable to input some CSV properly. That's brilliant engineering right there. When my mother types on her computer, she just wants things to work. When she searches for something, she doesn't care if she typed the wrong incorrect Unicode character. That's the cases that need to be solved for users. And it is only O(2ⁿ) algorithm if you naively look at text as an array of bytes. Time to, perhaps, broaden some horizons.