Subversion Repositories filter_foundry

Rev

Rev 365 | Rev 369 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
220 daniel-mar 1
 
2
Wishlist/ToDo-List
3
==================
4
 
274 daniel-mar 5
Known bugs
6
----------
220 daniel-mar 7
 
366 daniel-mar 8
(None)
220 daniel-mar 9
 
249 daniel-mar 10
 
220 daniel-mar 11
Minor priority stuff or ideas
12
-----------------------------
13
 
366 daniel-mar 14
* Add functionality to save to FFX (Filters Unlimited) and PFF (Premiere Filter Factory) format
15
 
365 daniel-mar 16
* .gitattributes eol=cr => -text or binary
17
 
294 daniel-mar 18
* Filter Factory is much faster than Filter Foundry (example rad.afs). Why?!
19
 
276 daniel-mar 20
* When a filter is created obfuscated and you click "Make" again, should then be the "obfuscate" checkbox be checked again?
21
 
228 daniel-mar 22
* The filter `r*(y&1)` looks horrible when you zoom out!
221 daniel-mar 23
 
228 daniel-mar 24
* Memory leak: `strdup()` and `my_strdup()` need `free()` !
220 daniel-mar 25
 
26
* Why can't we edit *.rc files in Visual Studio?
27
 
346 daniel-mar 28
* `host_preserves_parameters` (enabled with GIMP/PSPI) should somehow delete the temporary AFS file at each restart of GIMP. Otherwise, the user would always see the previous session when they re-open GIMP.
220 daniel-mar 29
 
228 daniel-mar 30
* Create and evaluate more testcases, and either fix differences between FilterFactory and FilterFoundry, or document them in "Filter Factory Compatibility.md"
220 daniel-mar 31
 
228 daniel-mar 32
* Should an expression like "----r" be forbidden? (Multiple negate signs) On the other hand, FilterFactory didn't forbit them either.
220 daniel-mar 33
 
228 daniel-mar 34
* CMYK mode is possible (although a bit misleading to have r=c, g=m, b=y, a=k), but then it is impossible to control the alpha channel.
220 daniel-mar 35
 
228 daniel-mar 36
* I have found following in the source code... Do we need to do something here?
220 daniel-mar 37
 
228 daniel-mar 38
        strcpy(gdata->parm.formula[i],expr[i] ? expr[i] : "bug! see builddlgitem");
39
 
220 daniel-mar 40
* Search for "TODO", "FIXME" and "Codereview" in the code
41
 
42
* There is no warning if a formula contains a number that exceeds 32 bits.
43
 
228 daniel-mar 44
* Minor bug: Testcase testcases/rst_3.afs applied to a 1000x1000 canvas: When the preview is zoomed in to 29% or 59%, and the preview is panned, the bars change during panning. It does not look "smooth" like in 100%, 50%, or 25% zoom. The problem is that the offset of the preview area is always different, and if the zoom level is not a multiple of two, you will always "pick" other bars.
45
 
220 daniel-mar 46
* Fast (double) click in [+]/[-] scroll buttons is not accepted as 2 clicks / zoom-requests
47
 
48
* Support more colors modes and 16bit. Why is Lab color not accepted, although doesSupportLABColor is set?
228 daniel-mar 49
 
220 daniel-mar 50
* Add Unicode support? (Does PiPL support Unicode at all?)
51
 
52
* Picture with 78x63 in grayscale mode: In preview on the left side there is a black bar
53
 
346 daniel-mar 54
* With the change in 1.7 in re "*preview evaluates whole picture if state changing functions rst,put,rnd are used*": Is there anything we can do to make it more performant when the user changes sliders? Especially with large pictures, the process becomes a bit laggy! => This feature was removed in 1.7.0.9, because it was extremely slow for some filters!
228 daniel-mar 55
 
220 daniel-mar 56
* Check why the plugin crashes Premiere 5 (called in image filter mode; obviously not as transition filter)
57
 
311 daniel-mar 58
* Cosmetics: Include a plugin-like-icon as icon #1 so that there would be a good icon in case some apps try to display an icon from the 8BF file?
59
(Attention: make_win.c deletes icon #1 because it thinks that it is the exclamation icon, so this needs to be changed then)
220 daniel-mar 60
 
346 daniel-mar 61
* Let PIPL have resource ID 16 instead of 16000, so that other apps might be able to recognize it as Filter Factory plugin?
311 daniel-mar 62
 
346 daniel-mar 63
* Make Filter Foundry ready for translations? In Windows, put all strings in string lists (resources, `LoadStringA`), as well as in Mac resources.
64
 
65
* Should the compiler flags in `funcs.h` placed as resource (binary bits), so that the behavior can be changed if required?
66
 
67
 
366 daniel-mar 68
Big ideas
69
---------
70
 
71
* How about implementing the "Filters Unlimited" (FFX) language?
72
 
73
 
303 daniel-mar 74
Questions
75
---------
76
 
346 daniel-mar 77
* What is the exact purpose of the variable `needinput`?
78
  It is set if we use rgbaiuvc or `cnv()`
79
  But why isn't it set if we use `src()` or `rad()`. Is that OK???
312 daniel-mar 80
 
346 daniel-mar 81
* Why does everybody write that the Filter Factory `tan(x)` function is bounded?
303 daniel-mar 82
  The Filter Factory manual writes that it is bounded, and the Filter Factory Programming Guide
83
  does also mention it (including a graphic).
84
  Is it a bug that it goes infinity? Do we need to fix it in Filter Foundry???
85
 
86
  Following examples of Filter Factory show that it is going off-bound! (Also tested with PS 3.0.5)
87
 
88
    tan(255)  ==  167761 ? 255 : 0
89
    tan(256)  == -167772 ? 255 : 0
90
    tan(-768) ==  167772 ? 255 : 0
91
 
335 daniel-mar 92
* Is `D` is supposed to be a synonym `dmax` (then `D` should stay 512),
93
  or is `D` supposed to be `dmax-dmin` (then `D` should be 1024 and `dmin` should be 512)?
303 daniel-mar 94
 
353 daniel-mar 95
* In regards length of the PIPL TLV, should it contain padding or not? Asked in https://community.adobe.com/t5/photoshop-ecosystem-discussions/question-about-pipl-tlv-encoding/m-p/12377822
346 daniel-mar 96
   - No:
97
	* Photoshop does not accept if padding is not added to the length value
353 daniel-mar 98
	* pluginsdk\photoshopapi\pica_sp\SPPiPL.h states: "Number of characters in the data array. Rounded to a multiple of 4."
346 daniel-mar 99
	* Official Adobe plugins, e.g. „3D Transform.8bf“ are rounding the length to a multiple of 4 (actually, rounding to the next possible multiple 4, so that padding is always guaranteed).
353 daniel-mar 100
	* cnvtpipl.exe does the same
346 daniel-mar 101
   - Yes:
102
	* The 1997 PICA documentation (page 23) and 1996 "Cross-Application Plug-in Development Resource Guide" describe both: "Contains the length of the propertyData field. It does not include any padding bytes after propertyData to achieve four byte alignment. This field may be zero."
353 daniel-mar 103
	* pluginsdk/documentation/html/struct_p_i_property.html writes the same thing
335 daniel-mar 104
 
346 daniel-mar 105
  Is there really a defect in these documentations?
106
 
107
 
220 daniel-mar 108
Design/UI Tweaks
109
----------------
110
 
111
* Should it be possible to zoom more than 100%?
112
 
228 daniel-mar 113
* The vertical scrollbar should auto-hide when the expression formula is short. (Also for Mac). Unfortunately, this task is very hard in WIN32 API.
220 daniel-mar 114
 
343 daniel-mar 115
* Should there be more (visual) space for expressions, like in Filter Factory? Also, Filter Factory had more visual space for slider names (e.g. required in Alf's Power Tools).
220 daniel-mar 116
 
117
* Sliders are extremely tiny!
118
 
119
* With some canvas formats, the initial zoom level is not "fully zoomed out"
120
 
228 daniel-mar 121
* If you load an AFS file which has multi-line expressions, then the first expression (R) is scrolled down (because it has the focus)
220 daniel-mar 122
 
271 daniel-mar 123
 
124
Regarding Macintosh
125
-------------------
126
 
231 daniel-mar 127
* Unsure regarding AppleScript !
128
	https://developer.apple.com/library/archive/documentation/mac/pdf/Interapplication_Communication/AE_Term_Resources.pdf
129
	1. Do we need an 'auet' resource?
130
	2. Does our implementation work with AppleScript?
131
	3. Must the human-language term REALLY be unique?! So only one plugin on the whole system may use the word "ctl(0)" or "red" or "size"??? Why do we have unique keys if the human language needs to be unique too?!  (see "Important" section at page 8-14)
228 daniel-mar 132
 
133
* In ui_mac.c, there does not seem to be a limit for the expressions. So, is it possible to write an expression longer than 1023 bytes, which would lead to a crash?
220 daniel-mar 134
 
228 daniel-mar 135
* Macintosh: Please look at the (*) marked entries in CHANGELOG.md and verify the bugs/fixes and/or apply the changes on Mac.
220 daniel-mar 136
 
228 daniel-mar 137
* Macintosh: Is it possible to create a 64 bit version of FilterFoundry? Do we need to rewrite the UI code because the Carbon Framework is replaced by Cocoa!
220 daniel-mar 138
 
139
* Some *.r files include "PiMI.r". Is this part of an old SDK or is a file missing?
140
 
346 daniel-mar 141
* Support BigDocument (Add "ms32" PiPL, set `pb->bigDocumentData->PluginUsing32BitCoordinates` and use `pb->bigDocumentData->*`)
220 daniel-mar 142
 
347 daniel-mar 143
* Correctly implement obfuscation version 6 (need to fix the binary executable code during build!)
232 daniel-mar 144
 
346 daniel-mar 145
 
220 daniel-mar 146
Things to check/correct for other Telegraphics plugins (FIXED in FilterFoundry)
147
------------------------------------------------------
148
 
228 daniel-mar 149
* The lengths of the PIPL TLV structure must include everything, including string paddings, with 1 zero padding guaranteed. See annotations in "GIMP Incompatibilities.md".
220 daniel-mar 150
 
228 daniel-mar 151
* The language/country code in VarFileInfo must be a code that exists in StringFileInfo. The translation strings "04900000" don't fit together "080904B0".
220 daniel-mar 152
 
153
* 64 Bit Windows builds
154
 
155
* XP Manifests (+ activation contexts)
230 daniel-mar 156
 
157
* Fix AETE resources, make keys unique and remove scope in order to enable AppleScript
232 daniel-mar 158
 
346 daniel-mar 159
* Support BigDocument (Add "ms32" PiPL, set `pb->bigDocumentData->PluginUsing32BitCoordinates` and use `pb->bigDocumentData->*`)
303 daniel-mar 160
 
161
* Place plugin dialogs on the correct screen