On Wed, 21 Nov 2018 09:30:39 +1100, Andrew Rowley wrote: >On 21/11/2018 5:26 AM, Phil Smith III wrote: >> The funny part about case sensitivity is that if you ask a *ix person why >> it’s good, they almost universally assert that it is, but cannot come up >> with a reason why, OR a case where you would deliberately mix two files or >> commands with the same letters but different case (“CONFIG.txt” and >> “config.txt”, et sim). >> Not only filenames, but names of other objects. For example, in HTML charcter entities, "é" means "é; "É" means "É".
>> I’ve also always been surprised that no *ix implementation ever bit the >> bullet and tried to fix case sensitivity. Windows, of course, got it right; >> alas, given the historical antipathy *ix folks have for Windoze, I fear >> that’s all the more reason it will never get fixed… >> "Right/wrong" and "broken/fixed" are highly subjective. NTFS is case-sensitive (perhaps for servers to accommodate UNIX clients). There's a Registry setting telling applications whether to use the case-sensitive or case-insensitive filesystem interfaces. Most applications ignore this; Cygwin respects it; I've tried it. It's startling to see "foo", "Foo", and "FOO" all in the same Explorer window. Even more startling that when I click on one, it's unpredictable which opens. That should be fixed. >I think one of the problems is the difficulty of implementing case >insensitivity in some other languages. English is straightforward with a >1:1 mapping between upper and lower case. I believe that is not the case >in all languages. > >So then you have the problem that translating to upper case and >comparing can give a different answer to translating to lowercase and >comparing. To implement case insensitivity you need to actually define >how a case insensitive comparison should be done. Then, does that >definition need to cover all languages, or would you allow a compiler to >accept or reject a program based on locale? Or all programs must be >written in English? > To be case-insensitive in English, Cyrillic, Greek, ... a filesystem would need to embed Unicode Services -- a peformance drag. Don't be Anglocentric. >At some point the easiest conclusion is that different is different, >which implies case sensitivity. > +1 Some have argued that a filesystem should resolve ligatures: not all of "Amoeba", "amoeba", "amœba", and "ameba" should differ. But which should be treated as equivalent? And diacriticals. Some have argued that they should "not matter" in filenames. But a Spanish speaker would almost certainly insist that "Mi_maravilloso_año" and "Mi_maravilloso_ano" do not identify the same file. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
