Optimize decoding tree to save more memory #2

New Issue

2020-12-31T08:05:22-05:00

tcsullivan commented

2020-12-31 08:05:22 -05:00

(Migrated from github.com)

Most importantly, we need to allow for larger decoding trees. Secondly though (and maybe this should be a separate issue), the decoding tree could likely be refined/redesigned to either take up less memory or work more efficiently.

See [here](https://www.reddit.com/r/cpp/comments/kn1wpe/compiletime_string_compression_using_huffman/ghjlj02/). Most importantly, we need to allow for larger decoding trees. Secondly though (and maybe this should be a separate issue), the decoding tree could likely be refined/redesigned to either take up less memory or work more efficiently.

tcsullivan commented

2020-12-31 09:54:43 -05:00

(Migrated from github.com)

Forgot that the input data gets left uncompressed if compression doesn't save space. The bug was simply in decoder::end returning a byte beyond what it should have for the uncompressed scenario. I'll commit the fix for the bug, but will still leave this issue up for the possibility of enhancing the decoder tree.

Forgot that the input data gets left uncompressed if compression doesn't save space. The bug was simply in `decoder::end` returning a byte beyond what it should have for the uncompressed scenario. I'll commit the fix for the bug, but will still leave this issue up for the possibility of enhancing the decoder tree.

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: clyne/consteval-huffman#2