[Toybox] [PATCH] vi: Replace linelist with mem_block based design

[Rob Landley]

On 1/16/20 7:58 AM, Jarno Mäkipää wrote:
> Replaced dlist linelist with continuous memory blocks. This will allow
> editing huge files without billion mallocs.

More or less what I was trying to implement back when vi was a thing I was going
to work on.

> File is first opened with
> mmap() and mapped region is fully described in block_list as one block.

When I tried this it turned into a complex mess of different objects with
different allocation/lifetime rules that had to interact. I intended to cap the
block size because splitting becomes painful otherwise (and then wound up having
strings individually allocated with the array of pointers to the start of
strings being a seperate allocation), and of course mmap() requiring the offset
to be a multiple of page size combined badly with line starts being all over the
place...

(I never did figure out what to do about a file that changed out from under you
while you were editing it. The MAP_SHARED version has the changes invalidate
your indexes and segfault, especially if it's truncated out from under you. The
MAP_PRIVATE version doesn't work on nommu, I believe still has the truncate
problem, and doesn't help you with parts it hasn't faulted in until after the
file changed.)

> Currently "valid" data is described as slices, when first loading file
> there is only one slice that points to memory existing in block_list.
> When cutting text, block_list is not freed or modified, but instead
> slice_list is modified to have "hole" between 2 slices. when inserting
> new mem_block is added, previos slices are cut in cursor position and
> new slice is added...

*shrug* I leave this to you.

> Implementation is not as finished as old linelist, but most of the heavy
> lifting is done so wanted to share this at this state.
> 
> Added functions to handling data inside block_list+slice_list
> 
> insert_str(), cut_str() are used for all delete and add operations
> text_strrchr(), text_strchr() are used for searching lineendings
> text_byte(), text_codepoint(), text_getline() are for simple data access
> 
> Implemented: yank,delete,insert,push and most of the previos moves
> 
> Implemented partly: saving of file, since current file is still mmaped
> cant save on top of it,

You never want to do that anyway because a crash halfway through (battery dies,
etc) leaves you a corrupted file.

> so now saves into "filename".swp, later need
> to move .swp into original file after munmap

rename() is atomic.

> Not Implemented: search, linenumber counting, etc..
> 
> Removed: Some unused functions
> 
> br
> 
> -Jarno

Do I apply this or not?

Rob