Current COW strings to be replaced with immutable version.

[bacek]

This will improve performance, reduce codebase, etc.

Various "inplace" string modification functions to be removed.

[bacek]

Also all Parrot_str_foo functions that have AGRMOD(STRING *dest) output parameter will be changed.

[bacek]

Also 2-args inplace forms of string ops are deprecated. E.g. chopn, etc.

[coke]

See r45730 : -1 from me.

I would recommend that instead of picking a new name, that we just refer to this as substr, but with a different signature.

[bacek]

Coke, why?

"substr" produce substring of original string. "replace" produce string with replaced part.

-- Bacek

[bacek]

We do need separate replace_s_s_i_i_c op because current substr_s_s_i_i_c has different (and bad) semantic.

-- Bacek

[coke]

bacek - Ok, +0 from me. =-)

[coke]

Bacek - can we get an explicit list of what is left to do on this ticket?

[coke]

Only inout string opcode left seems to be concat_s_s. (pin/unpin are not covered by this ticket.)

[coke]

Note that just removing this opcode is insufficient. ".=" in PIR desugars to this, so we need to fix this to use a suitable remaining opcode.

[coke]

This ticket also covers the inplace 'reverse' method on String. An acceptable fix is to make this an instance method that operates on the String itself instead of a passed in string.

[julian.notfound@…]

>  This ticket also covers the inplace 'reverse' method on String. An
>  acceptable fix is to make this an instance method that operates on the
>  String itself instead of a passed in string.

The 'trans' method also modifies in place its argument.


-- 
Salu2

[coke]

assign to plobsing for remaining IMCC work.

[plobsing]

Replying to coke:

assign to plobsing for remaining IMCC work.

What does changing opcodes which mutate registers, which are *pointers* to strings, not the strings themselves, have to do with COW vs immutable strings? This seems like a red herring to me.

[plobsing]

inplace concat IMCC patch

[plobsing]

Replying to coke:

assign to plobsing for remaining IMCC work.

I've attached a patch for the IMCC work. Still not really sure this is worthwhile.

[bacek]

Replying to plobsing:

I've attached a patch for the IMCC work. Still not really sure this is worthwhile.

+1 for applying patch.

-- Bacek

[bacek]

Hello.

Actually String PMC methods are _not_ covered (and shouldn't be). Ticket was about STRING*. So, after applying plobsing's patch and removing concat_s_s opcode ticket can be closed.

-- Bacek

[Vasily Chekalkin]

Remove concat_s_s op and rebootstrap generated files. Closes #1540

Changeset: ddbdd541c6cf9d1d9deb2a4d589bf1fcae416b12