Re: problem with applying md5 to data
Re: problem with applying md5 to data
- Subject: Re: problem with applying md5 to data
- From: Wilker <email@hidden>
- Date: Wed, 29 Jun 2011 10:31:45 -0300
Thanks Quincey, these things for sure will be a lot helpful, Im still pretty
new to Objective-C, but I will look in all of those things that you said,
thanks :)
---
Wilker Lúcio
http://about.me/wilkerlucio/bio
Kajabi Consultant
+55 81 82556600
On Wed, Jun 29, 2011 at 5:13 AM, Quincey Morris <email@hidden
> wrote:
> On Jun 29, 2011, at 00:27, Quincey Morris wrote:
>
> >> [fileData appendData:[file readDataOfLength:CHUNK_SIZE]];
> >> [file seekToFileOffset:MAX(0, fileSize - CHUNK_SIZE)];
> >> [fileData appendData:[file readDataOfLength:CHUNK_SIZE]];
> >>
> >> [file closeFile];
> >
> > This is not the worst way to read the entire contents of a file into a
> NSData object, but it's pretty bad. :)
> >
> > It would be far easier and more efficient to use [NSData
> dataWithContentsOfFile: path options: NSDataReadingMapped |
> NSDataReadingUncached error: NULL];
>
> Er, two things.
>
> I didn't notice until after I posted that you weren't reading the whole
> file. So, sorry, "pretty bad" was too strong -- I'm upgrading my response to
> "not the best". :)
>
> I think the way I suggested is still the best way, because (typically)
> using the NSDataReadingMapped means the file contents are mapped into
> virtual memory, so pages (4KB, or whatever the VM page size is) are not
> actually read until you try to access them.
>
> If you use CC_MD5_Init/Update (first part)/Update (last part)/Final instead
> of the convenience method that does it all, only the parts of the file you
> need will actually be read.
>
> NSDataReadingMapped is in general the most efficient way to read a file
> when you only need to read any part once, I think.
>
> The problem with NSFileHandle is that it pretty much sucks as a class. It
> doesn't return any errors, so if something goes wrong it's either going to
> throw an exception or just ignore it. IIRC it has no API contract regarding
> error reporting.
>
> Second thing...
>
> I forgot to point out that your code has a glaring bug. With your
> declarations, 'fileSize - CHUNK_SIZE' in 'MAX(0, fileSize - CHUNK_SIZE)' is
> an unsigned expression. It can't go negative, and so that statement isn't
> doing what it looks like it's doing.
>
> One more thing ...
>
> Since you have no release/retain calls, I'm assuming you're using garbage
> collection. If that's the case, then this line is very, very dangerous:
>
> > return [SMEncodingUtilities md5HexDigestFromChar:[fileData bytes]
> withLength:(unsigned int)[fileData length]];
>
> You're passing an *interior* pointer to fileData's private storage, but the
> (compiler-optimized) lifetime of 'fileData' ends *before* the md5 method is
> entered, and so the object is subject to being collected *before* you've
> used the pointer. This will crash.
>
> In your code, the window of danger is pretty small. But it will crash
> eventually.
>
> The solution is to put a reference to the object after the call:
>
> > NSString* result = [SMEncodingUtilities
> md5HexDigestFromChar:[fileData bytes] withLength:(unsigned int)[fileData
> length]];
> > [fileData self];
> > return result;
>
> It's a PITA, but you gotta be careful with NSData.
>
>
_______________________________________________
Cocoa-dev mailing list (email@hidden)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden