|Version 13 (modified by nwellnhof, 6 years ago)|
Remove charset opcodes and fixed_8 encoding
Charsets have been merged into encodings. The charset opcodes and the fixed_8 encoding will be removed.
After the charset/encoding merge, they're unneeded.
Use the corresponding encoding opcodes instead.
/* old code */ $I0 = charset $S0 $S1 = charsetname $I0 $I1 = find_charset "ascii" $S2 = trans_charset $S0, $I1 /* updated code */ $I0 = encoding $S0 $S1 = encodingname $I0 $I1 = find_encoding "ascii" $S2 = trans_encoding $S0, $I1
The list of supported encodings is:
ascii iso-8859-1 binary utf8 utf16 ucs2 ucs4
Use the 'utf8' encoding instead of the 'unicode' charset. The 'fixed_8' encoding is also going away. Use 'ascii' instead. If you want to test for a fixed_8 encoding, you have to compare the encoding to 'ascii', 'iso-8859-1' and 'binary' separately.
The C function Parrot_PCCINVOKE has been removed.
Use Parrot_pcc_invoke_method_from_c_arg instead. A simple textual substitution is enough to update any code that depends on this function.
/* old code */ Parrot_PCCINVOKE(interp, interp->scheduler, CONST_STRING(interp, "count_handlers"), "S->I", handler_type, &count); /* updated code */ Parrot_pcc_invoke_method_from_c_args(interp, interp->scheduler, CONST_STRING(interp, "count_handlers"), "S->I", handler_type, &count);
See r42129 for more examples.
Remove Parrot_find_global_s and Parrot_store_global_s
Parrot_find_global_s and Parrot_store_global_s were removed after they were found to be unused.
Nothing in Parrot used them and they were therefore untested.
Use the body of one of the old functions in place of its call.
/* old code */ XXX: insert code here /* updated code */ XXX: insert code here
See r48435 (src/namespace.c changes) for an example of what to replace calls to these functions with. See r42129 for more example
Change behavior of find_lex opcode
The find_lex opcode no longer throws an exception when the lexical is not found; it returns PMCNULL instead.
This is consistent with the other "lookup" opcodes.
If you want to throw an exception, you must check the result of the opcode and throw the exception explicitly.
Remove CodeString PMC
The CodeString pmc has been removed.
See #1633. With the switch to immutable strings, CodeString became very expensive, as it did a lot of string concatenations.
As a replacement for the .emit() functionality, instead use .append_format() in StringBuilder. Only difference from emit() is that newlines are not automatically appended, so you'll have to add them manually.
Charname_to_ord is now an opcode: find_codepoint. The other methods that were provided are now part of PCT (and are probably not needed directly).
Eliminate Raw NCI
It is no longer possible to create an NCI pmc without a corresponding NCI signature.
This functionality abused the NCI and went against the initial purpose - bridging the gap between parrot and C by translating calling conventions.
This was error prone, generally leading to segfaults.
If you must have an equivalent, take a look at NativePCCMethod. Otherwise, enjoy the fact that your segfaults are now most likely "not implemented" errors.
The function Parrot_get_runtime_prefix has been removed.
It returned a char * instead of a Parrot string. A new function provides better functionality and there is no need to keep this.
The method is_tty in several Handle PMCs has been removed.
There was a inconsistency, some PMCs used is_tty and others isatty. The decision was to keep isatty.