Functions isfile() and isdir() from std.file test if the name is a file or directory, but if the file or directory doesn't exist they throw an exception. I'd say it's not a file or directory if it doesn't even exist. Well, I don't mind the current behavior if it would specify this in phobos.html; and I suggest adding another function exists().


-- 
Christopher E. Miller

Ughh, are you serious it throws an exception ?

Can we please change this to return boolean ?

Phobos has become a seriously crippling element for the D language.  But im done preaching (begging?), ive shouted myself hoarse. What a shame.

Charles

On Fri, 7 May 2004 16:26:51 +0000 (UTC), Vathix <vathixSpamFix@dprogramming.com> wrote:

> Functions isfile() and isdir() from std.file test if the name is a file or
> directory, but if the file or directory doesn't exist they throw an exception.
> I'd say it's not a file or directory if it doesn't even exist. Well, I don't
> mind the current behavior if it would specify this in phobos.html; and I
> suggest adding another function exists().
>
>

C wrote:

> Ughh, are you serious it throws an exception ?
> 
> Can we please change this to return boolean ?
> 
> Phobos has become a seriously crippling element for the D language.  But im done preaching (begging?), ive shouted myself hoarse. What a shame.
> 
> Charles

I think this is the correct behaviour.  Asking the OS for information about something that isn't there at all sounds exceptional to me.

It's generally better for a program to explode spectacularly than to subtly fail in the face of oversight.  It's not at all unrealistic to suppose that someone might write something like this:

   if (isfile(fileName) {
      ....
   } else {
      // assume it is a directory
      ....
   }

 -- andy

"Vathix" <vathixSpamFix@dprogramming.com> wrote in message news:c7gdca$orb$1@digitaldaemon.com...
> Functions isfile() and isdir() from std.file test if the name is a file or
> directory, but if the file or directory doesn't exist they throw an
exception.
> I'd say it's not a file or directory if it doesn't even exist. Well, I
don't
> mind the current behavior if it would specify this in phobos.html; and I suggest adding another function exists().

I thought there was an exists() function already, but there isn't. You're right, there needs to be one.

I've attached the overlooked function.

Andy Friesen wrote:

> I think this is the correct behaviour.  Asking the OS for information about something that isn't there at all sounds exceptional to me.
> 
> It's generally better for a program to explode spectacularly than to subtly fail in the face of oversight.  It's not at all unrealistic to suppose that someone might write something like this:
> 
>    if (isfile(fileName) {
>       ....
>    } else {
>       // assume it is a directory
>       ....
>    }
> 
>  -- andy

Not always.  I am mentioning it again, and it's not even a compiled language, but PHP handles this differently for good reason.

Basically, because PHP supports streams (ssl://, http://, zlib://, ftp://, etc.) it's possible you could get your hands on a variable that contains a URL.  The is_file() function returns if this is a *regular filesystem* file... and, similarly, the is_link() function returns if it is a *regular filesystem* symbolic link. (whereas file_exists() might work over zlib:// and tell you if the file exists as a gzip, etc. or under ftp:// the normal way.)

But, assuming there is a function like exists(), or file_exists(), etc. it would not be a problem at all.

Either way, personally I see assuming that !file is dir as more of an oversight than using isfile on a non-existant path.

-[Unknown]

"Walter" <newshound@digitalmars.com> wrote in message news:c7gkv5$160c$1@digitaldaemon.com...
> I've attached the overlooked function.
>

You're returning false if getting the file attributes fails, but a different error could have occured like due to permissions or disk failure. I think it should test, on failure if GetLastError() returns ERROR_FILE_NOT_FOUND or ERROR_PATH_NOT_FOUND; and something with errno for linux, I assume. Sorry for nitpicking :>


--
Christopher E. Miller

Unknown W. Brackets wrote:

>>    if (isfile(fileName) {
>>       ....
>>    } else {
>>       // assume it is a directory
>>       ....
>>    } 
> Either way, personally I see assuming that !file is dir as more of an oversight than using isfile on a non-existant path.

That's exactly what I meant, actually.  If it isn't a bug, it sure looks like one to me.  If it is, then it would be preferable for the program to instantly explode with a line number and a good error message.

 -- andy

"Vathix" <vathixSpamFix@dprogramming.com> wrote in message news:c7gnga$19im$1@digitaldaemon.com...
> "Walter" <newshound@digitalmars.com> wrote in message news:c7gkv5$160c$1@digitaldaemon.com...
> > I've attached the overlooked function.
> >
>
> You're returning false if getting the file attributes fails, but a
different
> error could have occured like due to permissions or disk failure. I think
it
> should test, on failure if GetLastError() returns ERROR_FILE_NOT_FOUND or ERROR_PATH_NOT_FOUND; and something with errno for linux, I assume. Sorry for nitpicking :>

It's a good point, and I could argue it either way.

Vathix wrote:

> You're returning false if getting the file attributes fails, but a different
> error could have occured like due to permissions or disk failure. I think it
> should test, on failure if GetLastError() returns ERROR_FILE_NOT_FOUND or
> ERROR_PATH_NOT_FOUND; and something with errno for linux, I assume. Sorry
> for nitpicking :>
> 

Many other implementations simply say a file does not exist if you cannot access it too.  It's called security through obscurity :P. (kidding)

-[Unknown]