Hi Bart,

thanks for making SYS more elegant again and for re-introducing
the file timestamp copy function :-). Having separate readfile
and writefile makes the diff longer but the code cleaner :-).

About the exeflat thing: Makes the makefile simpler but makes
crosscompiling harder, if anybody was actually doing that...
I mean: We do not have many helper apps, but I guess of those
that we have, exeflat is one of the less portable ones...?
Just wondering. No real problem. Most people compile in DOS :-)

> It seems to work here now (SVN 1354).

Nice :-)

> I don't think shrinking was needed...

Then you should probably remove it - I think you are right, the
C library should already shrink, so what I do may actually waste
RAM by setting the size to 64k after the C library has set a
more suitable lower size before...

> that means that the compiler references all symbols with CAPITALS,
> unless you adorn them with _cdecl.

Confusing. Anyway. As long as it works...

> You should use "clobber" or the "-r" option with build that does the
> same thing. Without "-r" I got indeed a TGROUP>64k warning.

Whats the diff between clean and clobber and what does -r do?

> UPX's --lzma compressed kernels seem to crash

Remember that lzma uses a 2 pass decompressor which uses quite
a bit of RAM and which is 386+ only...

> FAT32/86 uncompressed NRV (LZMA)
> OW 1.6    69502     45841 (44206)
> OW 1.7    69165     45788 (44109)

Interesting that OW 1.7 saves 340 uncompressed but 50 compressed.

> FAT32/386 uncompressed NRV (LZMA)
> OW 1.6      67092    45565 (43967)
> OW 1.7      66531    45420 (43769)

Interesting that 386 saves 2.5k uncompressed and < 370 bytes UPXed.
So... Basically OW 1.7 386 really helps with RAM usage, but all
other variants are very similar in file size.

> For .sys compression (possible when uncompressed < 64K) it doesn't:
> FAT16/386(OW1.7) 62651 40921 (40921)

Hmmm well that is a nice size but what was the size with non-sys UPX?
This is 4k below FAT32 uncompressed, not bad. AND 4.5k saved UPXed :-)

Eric

PS: I guess the LZMA stub memory model just does not support .sys...
Or maybe UPX thinks the kernel is a dual exe/sys? But then you seem
to imply that you only use sys compression for non-exe-ish kernels?



-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/
_______________________________________________
Freedos-devel mailing list
Freedos-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freedos-devel

Reply via email to