Cache scheme

[merces]

I'm kinda proud of the cache scheme I implemented at

dz6/src/reader.rs

Line 10 in 918d9a4

pub struct Reader {

which I mainly use at

dz6/src/global/goto.rs

Line 27 in 918d9a4

if offset > self.reader.cache_offset_last {

but I suspect there's an easier way using BufRead. Would it be more suitable? The idea is that I can't load the whole file in memory at once as this slows down everything. Also, I have to decide how much to render at once (currently 1024 bytes). Does that need change?

[sergiogarciadev]

I am currently trying a new approach for this which uses MemoryMapped files. This will make the fix for #5 a lot easier and will let us implement things like backward search, and it also should be easier to handle and more efficient.

Also, the ideal amount of data to render should be dynamic, based on frame.area().height and app.config.hex_mode_bytes_per_line, this will avoid empty spaces in bottom of the frame when the terminal size is to big for the displayed data.

Having the file mapped on memory, will lets us handle it like it is an array but will efficiently read only the pages we access, making the hard codes values for cache and page size deprecated.

I will send a PR soon for you test and discuss.

[merces]

Thanks for your work. I'll take a look at the code next weekend. 😉