SDCC 4.5.0 Release

Delivery	Planned date	Actual date
SDCC 4.5.0 RC1
SDCC 4.5.0 Release

See the [SDCC 4.5.0 Release Plan].

Tasks

If you think there is a functionality that should be included or a bug that needs to be fixed in the release, please submit it to the Feature Requests at https://sourceforge.net/p/sdcc/feature-requests/ or Bugs at https://sourceforge.net/p/sdcc/bugs/ and copy Request ID and Summary fields to the following table, where:

Request ID: Request ID from Feature Requests or Bugs
Summary: Summary from Feature Requests or Bugs
Proposed by: Who proposed the feature or bug fix
Time: Estimate of the implementation time
Developer: Who will implement the feature or bug fix
Status: Current status

Request ID	Summary	Proposed by	Time	Developer	Status
[feature-requests:#279]	Accessing the stack	spth			work in progress
[Assemblers and Linkers]	sdas merge with asxxxx 5.0	borutr	a lot more	maarten, basxto, (borutr)	work in progress
[feature-requests:#324]	float classification macros	spth
[feature-requests:#380]	Handle static vars like global ones in initialization	spth
[feature-requests:#327]	mcs51 code in device/lib	spth
[feature-requests:#332]	Compound literals	spth
[feature-requests:#403]	Regression testing for pic16	spth
[feature-requests:#450]	K & R function style	spth
[feature-requests:#531]	Algebraic optimization on masked xor	spth
[feature-requests:#530]	Narrowing loop counters	spth			work in progress
[feature-requests:#543]	Leaner source: Remove pic14/pic16 autogenerated header and lib from source tree	spth
[feature-requests:#253]	Register parameter passing	spth		spth	work in progress
[bugs:#2700]	Debug info for local variables that live in multiple iTemps	spth
[bugs:#2701]	dwarf debugging: local variables has wrong scope	spth
[bugs:#2733]	False "overflow in implicit constant conversion" warning	spth
[bugs:#2906]	static local variables problem	spth
[bugs:#3303]	Wrong paths in the generated elf file	roybaer		roybaer	work in progress
[bugs:#3491]	Fresh build with --disable-sdbinutils will fail	spth
[bugs:#3555]	sdbinutils install doesn't respect prefix	spth
[bugs:#3765]	zdasz80 wrong compilation CP and LD commands	spth
[feature-requests:#937]	Code size regression due to recent type bugfix	spth
[bugs:#3770]	GCC 14 / clang 19: lots of host regression test failures	spth		@spth	DONE
[bugs:#3714]	SDCC 4.4.0 on Windows: Sdar tool crashes because of missing DLLs	spth
[bugs:#3686]	Snapshot doesn't run, misses DLLs	spth
[bugs:#3616]	Error for sdcc calling sdcpp/etc on WIndows non-C drives with path spaces	spth
[bugs:#3610]	Windows specific 32-bit sdar failure with modified CSEG	spth
[feature-requests:#567]	Eliminate unused string literals	spth		@spth	DONE
[bugs:#3788]	Incorrect optimization in bitfields / union combination	spth

Feature List

There is always a problem to create a list of new features when a new SDCC release is announced. The following list is an attempt to create such a list "on the fly". Every time a new feature which is visible or interesting for the SDCC users is implemented, it should be added to the following list.

• Full atomic_flag support for msc51 and ds390 ports.
• Experimental f8 port.

[Ragozini Arturo]

Request ID: #913 Inefficient assignment between int/long/pointer/union/struct
Request ID: #914 Allow isaddr flag not just for POINTER_SET?
Summary:The compiler does not copy efficiently pointers, int or long. It does not use LDI or other instructions with indirect addressing mode. Passing the flag "isaddr" in later stages of the code generation would allow the use of the Z80 instructions with indirect addressing mode like LDI, LDD, add a, (hl), and a,(hl), sub a,(IX+n) ecc ecc.
Proposed by: Arturo Ragozini
Time: n.a.
Developer: Philipp Klaus Krause
Status:n.a.

[Philipp Klaus Krause]

Such feature feel more like "nice to have", than necessary for 4.5.0. My personal idea for 4.5.0 is for it to be mostly a better 4.4.0: fix bugs found in 4.4.0, and try to avoid regressions. In general, IMO, 4.4.0 was already a good release wrt. bugs, but we've had a few annoying Windows-specific issues, and half our user base in on Windows. That isaddr flag bit looks like it would be too much for me before 4.5.0.

[Ragozini Arturo]

ok, it is a more advanced feature for a release aimed to bug fixing... maybe in the next version
Anyway you could give a glance to my tickets about 16 bit shift/ rotations
it should improve the code for math
Another area where the code generation can improve a lot is the use of push/pop instead of using the (ix) to access to the stack

[Ragozini Arturo]

Request ID: #198
Summary: Add symbols to know at link time the segment number of a label in banked Roms
Proposed by: Arturo Ragozini
Time: n.a.
Developer: Philipp Klaus Krause
Status:n.a.