flat assembler
Message board for the users of flat assembler.
Index
> IDE Development > "Fresh" work version with code completition Goto page 1, 2, 3, 4 Next |
Author |
|
JohnFound 08 Jan 2004, 20:03
It is available on the Fresh site. It is preliminary and the behaviour (and speed) are not good, but some opinions whould be usefull.
Regards. |
|||
08 Jan 2004, 20:03 |
|
Betov 08 Jan 2004, 23:57
Two words of explanations about what to do for testing would help
Betov. |
|||
08 Jan 2004, 23:57 |
|
Betov 09 Jan 2004, 09:13
Good work John.
I did not succeed to produce serious problems with CC. Just, i think it should close when the user input produces any modification of the Caret Pos not fitting with the going on Completion (for example Horizontal ArrowKeys, and any Menu Option activation. Also, i don't know if you really whished it to work this way, but, when inputing "RECT.", the ListBox comes out without hitting [Ctrl][Space], when you hit the '.'. (I remember you recommending to me to only run the ListBox on user [Crtl]/[Space], what i agread with...). If Fresh is Minimized, the ListBox remains on the screen. Then, when restored, it does no more work. It should either be closed or kept fully alive. Don't you want the [Esc] key to close the ListBox? Other side remarks: * Fresh Installation is terrific. This is, IMHO a point you should fix (turn Click&Go) as soon as possible (not that much work, is it?) * Why is it so slow to recompile Fresh? What Source sizes? (i see less than 1/2 Mega...). Bye, Keep up with the good work. Betov. |
|||
09 Jan 2004, 09:13 |
|
roticv 09 Jan 2004, 10:28
I think it is more like close to 1MB.
|
|||
09 Jan 2004, 10:28 |
|
Betov 09 Jan 2004, 11:23
I just suppose it is not really a Sources Size problem, roticv. My own comparisons Tests of the various Assemblers indicated to me, in the past, that FASM was a bit faster than RosAsm. So... Unless the Multi-Passes of FASM itself would increase the Compile time in a "exponential" way, i was supposing this was something particular to Fresh, and i would be estonished if RosAm, that is far from fully optimized for speed, actually, would be faster than FASM above 1/2, 1 or 2 Mega(s) Sources.
(Difficult subject, these days, when swindlers like Randall and Hutch are claiming that MASM is the fastest one, while basing their claims on tests designed, on purpose, to produce "oriented" results ). Betov. |
|||
09 Jan 2004, 11:23 |
|
JohnFound 09 Jan 2004, 11:26
Hi Betov.
Betov wrote: I did not succeed to produce serious problems with CC. Just, i think it should close when the user input produces any modification of the Caret Pos not fitting with the going on Completion (for example Horizontal ArrowKeys, and any Menu Option activation. Agree. Quote: Also, i don't know if you really whished it to work this way, but, when inputing "RECT.", the ListBox comes out without hitting [Ctrl][Space] It is because if you enter RECT. (or any other label) the only correct input is some local label (field) name. Why not to open the listbox? Actually I don't know exactly is there some similar case in RosAsm syntax, but if have, it is the right behaviour IMHO. Quote: If Fresh is Minimized, the ListBox remains on the screen. ... Bugs, bugs, bugs... Quote: * Fresh Installation is terrific. This is, IMHO a point you should fix (turn Click&Go) as soon as possible (not that much work, is it?) Hm, what you exactly mean? You should only unpack and start. Fresh does not need any instalation. Quote: * Why is it so slow to recompile Fresh? What Source sizes? (i see less than 1/2 Mega...). It is because of CC. (Shame on me! ) This is work version and algorithms are simply horrible. I am working on it. The normal compilation Time (for example with 1.0.1B) is 25s on my K6-2 500. The size of source is approximately 1Mbyte. Thank you for the help. Regards Last edited by JohnFound on 09 Jan 2004, 12:36; edited 1 time in total |
|||
09 Jan 2004, 11:26 |
|
Betov 09 Jan 2004, 12:24
For the Install, well, simply, after unzipping, runing, loading a Project, Click on [Run],... it ask me where are the "External Debugger" (What for?), where is the "Include Directory" (why doesn't it look inside its own Directory, where it should be by default), and then the Compilation fails because of default Path not provided (why doesn't it search it from where it is?).
For a release like FASM (FASM alone), this is normal, but for a specific IDE like Fresh, this is not, IMO. Having the IDE doing this dark work silently would also save you from having to answer to beginners asking you how to install. For the speed, yes, this is what i was supposing. Though not that much interresting, once the speed is enough for confortable dev of uge Apps, i will have to redo my comparison tests on 1Mo Sources. It seems to me that RosAsm should be a little bit faster than this, at 500 Mhz... and quite frankely... even a "very little bit" would make me naively happy. I will redo these tests when you will have released the next fixed version. Bye. Good work. Betov. |
|||
09 Jan 2004, 12:24 |
|
Kain 09 Jan 2004, 15:47
Hi Betov.
Unless you made any changes, I seem to remember that RosAsm asks the user to configure and locate certain files on first install. Is this not similar to FRESH asking for path locations? |
|||
09 Jan 2004, 15:47 |
|
Betov 09 Jan 2004, 16:08
Not exactely, Kain. In the old days, it simply required absolutely nothing and so forth asked for nothing at all. Then when we (Guga, mainly) did all of this work around the OS Data, we had to release the files separatly (in order to save user from having to reload all of this stuff for nop, at each update). Now, when users download the complete Package, the very first run searches inside its own Directory, finds out by itself where the required Files are, and configures automatically and silently. Of course, if all of this is done "by hand", and if the user unzips in unexpected Folders, the automatic process will fail pathetically, but not up to the point of not runing a full talking Help to explain to the user what is wrong and what he exactely has to do.
Nevertheless i am convinced that Fresh should do something similar. Not much work to implement, and much more confort for true beginners who most often have not even an idea about what an environement thingie could ever be. After all, the purpose of an IDE like Fresh, is, exactely, to save the users form such zero interrest tasks. When the user first Clicks on a [Compile] Item, this to _Compile_... Not to get a frustrating error Message. Betov. |
|||
09 Jan 2004, 16:08 |
|
decard 09 Jan 2004, 16:13
I must say you've done a great work, John! CC works as it should. Just the CC list isn't sorted alphabetically.
Only I don't want to try to compile Fresh on my P266 . |
|||
09 Jan 2004, 16:13 |
|
pelaillo 09 Jan 2004, 16:19
Betov wrote: ...(why doesn't it look inside its own Directory, where it should be by default), and then the Compilation fails because of default Path not provided (why doesn't it search it from where it is?). Yes, I have had the same concerning about this. Actually this work is one part of a command line parser I'm working on. At the end, Fresh will have that behavior. IMHO, all other paths should be required only when needed for the first time. Regards, |
|||
09 Jan 2004, 16:19 |
|
JohnFound 09 Jan 2004, 17:07
decard wrote: I must say you've done a great work, John! CC works as it should. Well, not exactly, but it will. Also, sorting. (Guess for what I need this "qsort.asm" in the libs directory. Quote: Only I don't want to try to compile Fresh on my P266 . It will be fixed. Even maybe this night. Regards |
|||
09 Jan 2004, 17:07 |
|
Betov 09 Jan 2004, 17:54
What "qsort.asm" John. You just have to set the LBS_SORT Style for the ListBox.
Betov. |
|||
09 Jan 2004, 17:54 |
|
JohnFound 09 Jan 2004, 18:19
Betov wrote: What "qsort.asm" John. You just have to set the LBS_SORT Style for the ListBox. The labels tree should be sorted by value of the labels, not by name. Also the sorted labels tree allows binary search. This will be useful in the debuger. Regards |
|||
09 Jan 2004, 18:19 |
|
Betov 09 Jan 2004, 18:41
In my opinion, you'd better consider these things completely separated ones. (Hello bugs, if not...).
I suppose you will want to have some Run-Time Source-Level Debugging, like in RosAsm. If true, there is no problem with time saving for this functionality. It will be much fast enough, anyway, as it will have to point out only _one_ single location (If understand what you are thinking about...). I see the point of not having the additional problem i have, with RosAsm being NOT case sensitive, so that you are tempted of doing the work once for all. But this choice would much probably produce more problems afterward than a specific approach for each implementation. Betov. |
|||
09 Jan 2004, 18:41 |
|
Tommy 09 Jan 2004, 19:30
Good start, John! I'm busy with football this week-end, but I'll see if I get some time to do some programming too...
Keep up the good work! Cheers, Tommy |
|||
09 Jan 2004, 19:30 |
|
JohnFound 10 Jan 2004, 20:22
Well, the fixed work version is on the site. Now the compilation is faster (it will be fixed more) and some bugs (caused big memory leaks) are fixed. There are small changes in behaviour too.
Please make some tests and post some feedback. Please test the speed of the compilation of "Fresh.fpr" file and report the time and what computer you are using: On my K6-2 500MHz - Fresh.fpr needs 20..21 seconds for compilation. Regards |
|||
10 Jan 2004, 20:22 |
|
Tommy 10 Jan 2004, 20:38
Fresh.fpr compiles in 7.941 seconds on my computer (1000 MHz)... The code completition is quite fast... There's one thing though: when I press Ctrl+Space the list does not appear (only after "<symbol>.")... Is this supposed to be?? IMO the list should apear when Ctrl+Space is pressed... Keep it up!
|
|||
10 Jan 2004, 20:38 |
|
JohnFound 10 Jan 2004, 20:52
Tommy wrote: when I press Ctrl+Space the list does not appear (only after "<symbol>.") The list apears only when there is some matching labels in the list. For example: You type: "A<ctrl>+<space>" this should list only labels begining with "a" or "A" characters. If there is no labels with "A" the CC window will be not opened. If your caret is on empty place, all labels will be listed. At least this is behaviour that I write and it works on my machine. If you have some other behaviour, it is probably a bug. Please give some more details. Regards. |
|||
10 Jan 2004, 20:52 |
|
Goto page 1, 2, 3, 4 Next < Last Thread | Next Thread > |
Forum Rules:
|
Copyright © 1999-2024, Tomasz Grysztar. Also on GitHub, YouTube.
Website powered by rwasa.