Вид для печати
dare you elaborate? This seems to me ok:Цитата:
Сообщение от NEO SPECTRUMAN
Where exactly Lua kills this approach?Код:MACRO callNoSp returnReg?, procedureAdr?
ld returnReg?,.returnAfterJp
jp procedureAdr?
.returnAfterJp:
ENDM
DUP 100
callNoSp ix,someProcedure
EDUP
someProcedure:
push bc
jp (ix)
странно на вид работает о_ОЦитата:
Сообщение от Ped7g
когда последний раз проверял макросы
по моему или неправильно находилась длина процедур и часть их потом затиралась
или jp вели на другие адреса
конечно еще нужно проверить в реальном коде
- - - Добавлено - - -
хотя это не выход из положенияЦитата:
Сообщение от Ped7g
мне нужна именно временная метка
так как
скомпилируется какКод:tamp_left_allocate_macro
callNoSp ix,someProcedure
end_tamp_left_allocate_macro
tamp_right_allocate_macro
callNoSp ix,someProcedure
end_tamp_right_allocate_macro
tamp_left_allocate_macro
jp $
end_tamp_left_allocate_macro
когда нужноКод:$8000 ld hl,$8006
jp procedure
$8006 jp $8006
$C300 ld hl,$C306
jp procedure
Код:tamp_left_allocate_macro
ld hl,temp
jp procedure
end_tamp_left_allocate_macro
tamp_right_allocate_macro
temp
ld hl,temp
jp procedure
end_tamp_right_allocate_macro
tamp_left_allocate_macro
temp
jp $
end_tamp_left_allocate_macro
Код:$8000 ld hl,$С300
jp procedure
$8006 jp $8006
$C300 ld hl,$8006
jp procedure
память может быть сильно фрагментированной
и такая конструкция работать не будет
- - - Добавлено - - -Код:dup 100
tamp_left_allocate_macro
callNoSp ix,someProcedure
end_tamp_left_allocate_macro
edup
наглядный пример
https://jpegshare.net/images/da/b7/d...dbf222475c.pngКод:org_allocate_macro $C010
di : halt
end_org_allocate_macro
org_allocate_macro $C021
di : halt
end_org_allocate_macro
DUP 100
tamp_left_allocate_macro
callNoSp ix,someProcedure
end_tamp_left_allocate_macro
EDUP
так же невозможно организовать выход из такой процедуры
так как последний ix должен вести на выход
That screenshot: so you did want "C019 ld ix,C023" to make "C00B" procedure return AFTER the "di:halt", right?Цитата:
Сообщение от NEO SPECTRUMAN
And the final "C046 ld ix,C04D" should be "ld ix,someExit"?
well, the second part is easy, just: ".100 callNoSp ix,proc" turns to ".99 callNoSp ix,proc" + "ld ix,exit : jp proc"
but the first part is changing the task completely, so you don't want to IX="after `jp proc`", but you need IX="ahead of next code block".
If I understand you correctly now.
даЦитата:
Сообщение от Ped7g
и этот кодовый блок может быть где угодно в памяти
what about this:Цитата:
Сообщение от NEO SPECTRUMAN
(took me couple of minutes of thinking to avoid Lua ... but in the end the extra helper macro did the trick :D )Код:; helper macro to make __COUNTER__ sub-word substitution possible (hiding it in macro arg)
; ("FollowingCode___COUNTER__" would not substitute, but "FollowingCode_suffix?" does)
MACRO nextFollowingCodeLabel suffix?
IFDEF FollowingCodeLabel
FollowingCodeLabel:
UNDEFINE FollowingCodeLabel
ENDIF
DEFINE FollowingCodeLabel FollowingCode_suffix?
ENDM
MACRO tamp_left_allocate_macro
; ... some allocation done and final ORG
nextOrg=(nextOrg+$6789)&$FFFF ; for fun jump around memory (not real "allocate")
ORG nextOrg ; where next code block should land
; mark this new code-block with "FollowingCodeLabel" label, and advance it for next code-block
nextFollowingCodeLabel __COUNTER__
; ... other preamble shared code ...
ENDM
MACRO end_tamp_left_allocate_macro
; ...
ENDM
MACRO callNoSp returnReg?, procedureAdr?
ld returnReg?,FollowingCodeLabel
jp procedureAdr?
ENDM
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; the memory location of code blocks is not important (could jump back and forth)
;
; the source code order of blocks is important - "FollowingCodeLabel:" is relative
; to where in source you currently are.
;
; And "tamp_left_allocate_macro" will advance it
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
nextOrg = $1234
DUP 100
tamp_left_allocate_macro
nop
callNoSp ix,someProcedure
end_tamp_left_allocate_macro
EDUP
; one manual block outside of DUP ...
nextFollowingCodeLabel __COUNTER__ ; define label for this block and advance it
daa ; manual block code
callNoSp ix,someProcedure
; exit-block manually added, just defining last FollowingCodeLabel label
FollowingCodeLabel: ; create one more "next code block" label for exit-block
; ... exit code (after last block)
di
halt
someProcedure:
; ...
jp ix
это внешние макросы про которые мы ничего не знаем и которые мы не трогаемЦитата:
Сообщение от Ped7g
это примерно как совет полезть редактировать C:\MinGW\include\stdio.h :)
:v2_dizzy_ok: (I don't see how that matters for that example, just extract the functionality and add it in the form valid for your use case - I can't produce your code, if you don't share it first, if you will re-read history of this topic, you keep coming with new bits of info after every response posting you working example fitting your previous rules - stops being fun after like third try, that you can't explain yourself clearly in the first post ;) .. so you have like one try left)
я написал 60+ уникальных меток рукамиЦитата:
Сообщение от Ped7g
до того как написать тут о том что временные метки в sjasm не работают как нужно :)
хотя я периодически сталкиваюсь с этой проблемой и хотелось бы иметь решение
но то что предложено выше
больше похоже на убегание от проблем чем на их решение
yep. I have no plans to extend temporary digit-labels to work with all instructions, the current design of making them valid only with `jr/jp/call` is fine by me (I think I understand *why* Sjoerd Mastijn designed it like that... even current sjasm 0.42 has still the same design of "reusable labels") (while checking sjasm docs, I see somewhat adjusted rules for regular labels, this looks interesting, the "!" prefix would be nice to add into sjasmplus - not related to your issue, just taking note for myself).
There are two issues about extending temporary/reusable labels:
1) I don't have vision of clearly-better solution, when I gave it some short thought, I see only more mess and confusion
2) I have lot of other stuff to do on sjasmplus (SLD export and warnings-system improvements, then release), but little time for it
And for me the proposed solution in my last answer with code looks good, like valid solution, feels to me much more solid than some temporary label hack.
(of course anyone else can redesign the feature and ask me (and other contributors) if we like the new design, and implement a patch and create pull request... it's open project and I would feel bad to just reject it)
If I will have nothing else to do on sjasmplus and feel bored, I will certainly revisit any remaining suggestions and try to pick some good ones.
With regard to your current issue, I provided you working solution. I don't see what more you want, sorry.
https://github.com/z00m128/sjasmplus...es/tag/v1.18.0
- some new/improved directives: WHILE<->ENDW, ELSEIF, DEFINE+, DISPLAY
- abs operator
- new warning system (with systematic suppression scheme)
- some bugfixes/implementation rewritten
- and more tiny details...
... I have a feeling this may be default setup of some guys here (but you may try first without and see if it's relevant)
Anyway, if you see some new warning with "warning[some_id]" form, then `-Wno-some_id` will switch it off completely, and line comment "; some_id-ok" will suppress it for particular line.Код:; ... in source ...
OPT -Wno-rdlow -Wno-out0 -Wno-trdext -Wno-trdext3
Have fun. :)