PRGMA:
unarchive pgrmFAKE
pgrmFAKE

It gives me an error, how do I do this correctly?

Thanks, this works great!

However, fake stays unarchived when I'm done, how do I archive it when I'm done with it?

Yeah, me too. I personally use Xeda's TPROG and everything I have is permanently archived once I finish developing them except for my DoorsCS games (because it unarchive/archives them anyways). I wish there was a shell that had the option to not allow program writeback, (or have it do so only if you do something after exiting,) supported everything, and had a homerun feature... But no, everything that's out there is lacking at least one of those things...

Anyways, the program I posted toggles the ability to unarchive AND archive. You can just do
archive prgmFAKE
Asm(prgmARCHIVEafter running it. I think it's best to toggle it off after using it; it might mess up your calc if it's active outside a program.

Test timings:
Garbage Collect 1: 0.98s
Garbage Collect 2: 0.38s
Garbage Collect 3: 0.43s  // so without anything to get rid of, it takes ~0.4s
Garbage Collect 4: 0.92s  // after running and exiting an archived 2068 byte BASIC program
Garbage Collect 5: 0.37s // back to ~0.4s
Garbage Collect 6: 0.81s //after running and exiting archived 3141 byte JUMP by ztrumpet
Garbage Collect 7: 0.44s // back to ~0.4s
Garbage Collect 8: 18.79s // After running and saving a game in 8063 byte Allecto, which does change

So:
I was felling quite suspicious that it didn't do that after the first 6 Garbage Collects, but after testing Allecto, I'm not sure.
Maybe DCS does do that and I was being paranoid with my human timings or maybe DCS doesn't and it just takes 21 times as long to garbage collect 8063 bytes than it does to garbage collect 2000 bytes... idk and I don't want wear out more of my flash ROM. lol I've been so cautious about that; I think I've only Unarchived/Archived something ~30 times since I got my calc.

edit: Someone should ask Kerm
edit2: Thanks, Deep. I can now keep my DCS games archived!

DoorsCS copies archived programs to RAM and copies back only if the program content changes (basically if it uses SMC).

That is what he said but all of my Axe programs get copied back after use, even those not using SMC. Now I use zStart and it does the job perfectly. It is even faster to launch programs too.

You might want to talk to Kerm about adding that option, MirageOS had it, and i can't imagine it would be difficult to implement (basically just don't call the code to copy the program back into archive). I don't know how Axe programs are handled, but how do you know none of the routines use SMC? Or is there a compile-time option to use only non-SMC routines? Many routines use SMC because it's generally faster and smaller than loading content from locations in SafeRAM, and for some things it's hard to imagine how to go about it without using SMC.

This has already been discussed on Cemetech on a topic called "My opinion/request for DoorsCS" or something like that. From what I remember of the discussion, Kerm says that DoorsCS doesn't writeback if it is not necessary, and Quigibo says that not any of Axe routines use SMC.
I think the problem comes from DCS and not Axe because I don't have the problem when I replace DCS with zStart.

Quote from: chickendude on October 29, 2012, 08:44:40 am

You might want to talk to Kerm about adding that option, MirageOS had it, and i can't imagine it would be difficult to implement (basically just don't call the code to copy the program back into archive). I don't know how Axe programs are handled, but how do you know none of the routines use SMC? Or is there a compile-time option to use only non-SMC routines? Many routines use SMC because it's generally faster and smaller than loading content from locations in SafeRAM, and for some things it's hard to imagine how to go about it without using SMC.

This has already been discussed on Cemetech on a topic called "My opinion/request for DoorsCS" or something like that. From what I remember of the discussion, Kerm says that DoorsCS doesn't writeback if it is not necessary, and Quigibo says that not any of Axe routines use SMC.
I think the problem comes from DCS and not Axe because I don't have the problem when I replace DCS with zStart.

Do you modify any of the data buffers you put in the program? That counts as "SMC" as far as the fact that you're modifying things within the program space.