www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - [Issue 8839] New: MmFile do not use Range

reply d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839

           Summary: MmFile do not use Range
           Product: D
           Version: D2
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Phobos
        AssignedTo: nobody puremagic.com
        ReportedBy: bioinfornatics gmail.com



13:51:31 PDT ---
Dear,
In first i explain why this module is important to me:

i would like use mmfile as in bioinformatic file are really huge and this
technology should be faster than classical way

But they are two problems:
- any example , documentation or whatever …
- MmFile do not use Range

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 17 2012
next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839


Jonathan M Davis <jmdavisProg gmx.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jmdavisProg gmx.com



PDT ---
It works just fine with ranges. Just slice it to get an array over the mapped
memory. I've done it before, and it worked great. Essentially, MmFile is acting
like a container rather than a range and therefore provides range access to
what it contains rather than being a range itself.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 17 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839




15:48:54 PDT ---
yes but it use opSlice method to do this not 
 - front back safe …
In more why it is a class instead a struct ? MmFile do not use inheritance

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 17 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839




PDT ---

 yes but it use opSlice method to do this not 
 - front back safe …


I don't know what you mean.


 In more why it is a class instead a struct ? MmFile do not use inheritance


It shouldn't be IMHO, but changing it would be a breaking change, so I wouldn't
expect it to be changed. I don't know why it was made a class in the first
place, but per the copyright, it looks like it came from D1 originally, and the
language could have changed quite a bit since MmFile was created, and it may be
that it made more sense at the time to make it a class, but I really don't
know.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 17 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839




15:31:12 PDT ---
I mean a Phobos range as
InputRange: requires the empty, front and popFront() member functions

more info here: http://ddili.org/ders/d.en/ranges.html

so a little question that is faster to 
 - read byte one by one
 - take a sub range by example 80 byte by 80 by using slicing

thanks

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 18 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839




PDT ---

 I mean a Phobos range as
 InputRange: requires the empty, front and popFront() member functions


Yeah. And arrays have that via std.array.empty, std.array,front, and
std.array.popFront. So, the array that you get from slicing MmFile is a fully
functional range. It's even random-access with slicing and length, so it
supports pretty much every range-based function there is. Making MmFile itself
a range wouldn't gain you anything. And in fact, making it a range would be a
_bad_ idea for the same reason that making any container a range is a bad idea
- iterating over it would consume it. You wouldn't want to do something like

auto mmFile = new MmFile("filename");
mmFile.popFront();

After that call to popFront, you would no longer have access to the first byte
of the file without creating a new MmFile, which is almost certainly not what
you want. Rather, what you really want is a range which contains all of the
bytes in the file and allows you to operate on them without affecting MmFile,
and slicing MmFile to get an array works great for that.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 18 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839




12:19:39 PDT ---
i totally agree with you now :-)

but one day it will good to switch Mmfile to struct that is not huge break

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 23 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839




16:05:44 PDT ---
Since slicing use void[] as return type instead ubyte you can't use UFCS you
need to cast it
length member should be maybe size_t i think

design this module should be a good thing.
If you break something is a minor break but you will get a better mmFile

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 30 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839




PDT ---
If you need to cast the slice to ubyte[], then cast it to ubyte[]. Given that
it's operating on memory, it's probably better that it use void[] rather than
ubyte[]. Regardless, given the increased resistance from Walter and Andrei to
any kind of breaking changes, breaking changes are going to need a very good
reason (and if anything, minor ones are more likely to be rejected, because you
get code breakage whether the change is large or small, and you generally get
less benefit from minor changes).

Given the chance, I'd definitely like to adjust a few things in MmFile, but I'm
not sure that we can at this point, and while some minor adjustments would be
nice, as far as I know, there's nothing seriously wrong with MmFile, so as nice
as some changes might be, they're not really needed.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 30 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839




16:26:08 PDT ---
about breaking change i understand is important to D to get an api "more
stable"

about ufcs:
-> do not works

_file[_currentPosition .. _file.length]
        .countUntil!( ( a ) => isWhite( cast(dchar) a ) );


-> works

countUntil!( ( a ) => isWhite( cast(dchar) a ) )
               ( cast(ubyte[]) _file[_currentPosition .. _file.length] );

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 30 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839




PDT ---
Then save the range as a variable and use UFCS on that, or this would probably
work:

(cast(ubyte[])_file[_currentPosition .. _file.length]).countUntil!(a =>
isWhite(a))();

I really don't see not being able to use UFCS on the slice directly as a big
deal.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 30 2012
prev sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=8839




16:40:25 PDT ---
for me that is ok but for new developer it could be little confused since in
first it will try the common ufcs approach. It will get a strange template
error about as void[] is used as return type. It will spent some time before
fond a way as in this special case he need to cast it.

So for me is ok but for others maybe add a little example with a ufcs approach
to help him ;-)

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 30 2012