fl.expr~

Alex Harker reporter

@stkr and @weefuzzy might also be interested so I'm tagging them.

So - there are a couple of libraries that I could use to make this happen. I have tried a couple of prototypes that work. However, the issue is that in both cases the resultant calculations are noticeably slower than using lots of objects. In the faster of the two cases the binary size is also large. Therefore:

How important is speed in making this useful, vs just the patching convenience of less objects?

One option is for me to manually parse the expression into a set of objects that are hosted under the hood (so basically you write the expression and it is identical to making a set of objects connected that way - I could improve things further by collapsing any constant expressions down to their result). That would be a tedious thing to have to do, but most likely it looks like it will beat an approach that is a generic parser - not exactly sure why, but I suspect the looping behaviour is better.

The reason not to do this would be that it is likely a massive pain to get working and a lot more work than the few hours I spent hooking up pre-existing code to an object. If, however the main goal is efficiency of patching then perhaps using a pre-existing library is viable?

2018-06-02T21:55:59+00:00

Comments (6)

Alex Harker reporter
- changed status to wontfix
Realistically I'm not going to write this object
- 2017-11-07T23:05:15+00:00
Alex Harker reporter
Issue ~~#49~~ was marked as a duplicate of this issue.
- 2018-01-29T23:51:08+00:00
Alex Harker reporter
- changed status to open
Opened due to renewed interest...
- 2018-01-29T23:51:33+00:00
Former user Account Deleted
Thanks
- 2018-01-30T14:33:18+00:00
Alex Harker reporter
@stkr and @weefuzzy might also be interested so I'm tagging them.

So - there are a couple of libraries that I could use to make this happen. I have tried a couple of prototypes that work. However, the issue is that in both cases the resultant calculations are noticeably slower than using lots of objects. In the faster of the two cases the binary size is also large. Therefore:

How important is speed in making this useful, vs just the patching convenience of less objects?

One option is for me to manually parse the expression into a set of objects that are hosted under the hood (so basically you write the expression and it is identical to making a set of objects connected that way - I could improve things further by collapsing any constant expressions down to their result). That would be a tedious thing to have to do, but most likely it looks like it will beat an approach that is a generic parser - not exactly sure why, but I suspect the looping behaviour is better.

The reason not to do this would be that it is likely a massive pain to get working and a lot more work than the few hours I spent hooking up pre-existing code to an object. If, however the main goal is efficiency of patching then perhaps using a pre-existing library is viable?
- 2018-06-02T21:55:59+00:00
Alex Harker reporter
- changed status to resolved
OK - well the results of this investigation (now complete) will be in the next build
- 2018-06-08T11:22:26+00:00
Log in to comment