Account to feasibility and size constraints

[mandriuska]

I just started to look into eopsin, and it's great! Will definitely contribute in the short future.

Meanwhile, when updated from 0.7.8 --> 0.7.10 I found that 'Wrapper' smart contract size went from 6kb to 20kb. Do you know why it happened? It would help me to understand the depths better.

[nielstron]

Might simply be because a lot of the stdlib that is unused was not optimized out. Or because it uses str(int) which is a rather large function. Need to improve the dead variable remover/run it several times to see whats going on.

UPDATE: I tried running the dead variable optimizer several times, without any improvements. I tried removing str(int), without any improvement. Will further investigate.

[nielstron]

I found the issue. Its the resolution of this issue: #43

Generally, assignments of names are not removed as they might throw a NameError. This leads to large parts of the prelude not being optimized away. Something like the following remains, even though neither bytes nor Value is ever accessed.

bytes = ... (injected builtin code)
TxId = bytes

[nielstron]

I will try to trim the optimizer to remove assignments that can not throw NameErrors (i.e. if they always occur in the same control flow after the name has been assigned)

[nielstron]

Version 0.7.11 should improve this (generated code went down to 15kb), but still 3x the previous size.

[nielstron]

A simple reason could also be that 0.7.8 was simply broken. The double function as redeemer/minter script was not correctly injected and the bytes constructor was missing. Still...

[nielstron]

plutonomy-cli is still able to squench the script size by 60% to 6kb, so still room for optimizations...

[nielstron]

Update: I found an issue when introducing the double function as spend/mint script that would duplicate the main function, effectively doubling script size.

Has been resolved in Dev now