freenode/#mezzano - IRC Chatlog
Search
18:57:23
fittestbits__
I looked through the ironclad code. Under #+unix, ironclad reads /dev/urandom. So we would want to implement something like /dev/urandom for mezzano.
18:58:23
fittestbits__
I'm not sure what that would take. On linux, it does use the entropy pool until it runs out, then uses a pseudo random number generator.
18:58:43
fittestbits__
I've never looked it to creating an entropy pool, so I have no idea what it takes.
19:08:38
froggey
there is a paper on how linux acquires and manages entropy, though I can't remember what it's called
20:53:58
fittestbits__
What's the command to upload a file to ix.io? I have a sample of a stream that I think does what you want.
20:55:47
fittestbits__
I just tried that and got "no results to fetch". So, no pointer to the file - maybe it failed?
20:59:08
fittestbits__
This is code for the "tag" file system I was playing around with on mezzano. The stream creates contains only a portion of the file - reads more when needed in the read case, writes portions in the write case as it needs more room in the buffer.
20:59:50
fittestbits__
Takes into account "seek" operations. I don't remember if it handles overwrite.
21:03:36
fittestbits__
Sure, but you have to handle the cases where only a portion of a sector is modified. So, for example, if you write just byte 0, you have to have read the at least a sector's worth of data
21:03:54
fittestbits__
so that when you writing it out, only the first byte of the sector gets overwritten.
21:06:57
fittestbits__
It's not, just another case that has to be handled everywhere. For example, when you seek (set position) you need to read enough around the new position so data that doesn't get overwritten
21:09:30
ebrasca
If you like to write-byte then if you have this sector you write to this sector if not you read it and write.
21:10:55
fittestbits__
I was more interested in the file system aspects: being backwards compatible with pathnames and directories etc without actually allocating any disk space for directories.
21:11:27
fittestbits__
So I wasn't worried about supporting every feature until I was sure I liked the overall structure.
21:13:26
fittestbits__
I hadn't thought about versions - I think that would be a higher level function.
21:15:52
fittestbits__
It's just my style of writing - leaving blank lines to help separate blocks of code - separate thoughts. Like using paragraphs in prose.
21:22:00
fittestbits__
yeah unfortunately, a lot functions in file systems and USB drivers tend to be too long - breaking them up into smaller pieces doesn't seem to make sense because the pieces are too specific
21:26:08
fittestbits__
To break up large blocks in to smaller more coherent chunks. Sorry it that makes it harder for you to read my code - when I modify your code, I'll stick to your conventions - no blank lines inside functions.
21:35:14
ebrasca
What do you think about :documentation like for example (partition :initarg :partition :documentation "Name of partition") ?
21:39:57
fittestbits__
I prefer to keep lines short - so I like using comments and, inside class definitions, the comments are usually on a previous line.
21:40:33
fittestbits__
I like short lines so that I can have multiple non-overlapping editor windows next to each other.
21:50:07
fittestbits__
Sure, I was wondering about that particular case - but in general there are places where I think more info is needed that just the variable name.
21:50:50
fittestbits__
The stream code I uploaded has several variables that I think are named OK, but need additional info about what exactly they mean.
21:51:14
fittestbits__
That's why the comments. But maybe the documentation feature would have been better.
21:59:47
ebrasca
I have leaned some lisp from this lisp style guide http://www.cs.umd.edu/~nau/cmsc421/norvig-lisp-style.pdf