history.txt

一些初级的网络编程

 8-03-99    Tuesday >
        Removing those 3 bytes from C connection part wasn't such a good
  idea, but 1 of them at least can be safely removed. Re-send the non
  perfect DEPPACK.

 9-03-99  > Wednesday
        Jibz spent 3 bytes to make DEPPTINY a bit safer (it was assuming
  that high part of ESP was always 0), and removed 6 more bytes from it.
  Now it's only 507, and better than previous ones. I think we could also
  spend one more byte to clear the direction flag, leaving it with 508.

11-03-99    Friday >
        Send the 508 one and this updated history till the entry before this
  one :-)

16-03-99    Tuesday
        Ported the last optimizations to my own C-less version of DEPACK16.
  Started to look again at DEPPACK. Let's start from old DEPPACK instead of
  growing it from DEPPTINY. I'm checking the good behaviour of the whole
  thing with every bit changed. It's a really frustrating job.

17-03-99    Wednesday
        I've isolated 3 test versions of DEPPACK, one of them detects the
  "bad infile" in the test file, another gives always the same result and
  the last one produces different results depending of previous memory
  state. Still I can't understand why it happens.

18-03-99    Thursday >
        Send the test files to Jibz. Let's see if we can find out what the
   hell is happening together.

21-03-99    Sunday
        I've finally find out a possible solution (with expensive 9 bytes
   fix) to detect bad infiles. Seems to work well. Ported it to the "grown
   from DEPPTINY" one.

22-03-99    Monday >
        Send the fix to Jibz.

24-03-99    Wednesday
        Ported the fix to DEPPTINY, removing another test to keep it under
  512 (actually 511). Found a bug in DEPPACK Jibz advised me about a while
  ago. Let's debug it once again. Fixed. Let's make a major test of
  everything. Ooopss!! First file tested: new bug found. This one crashes
  the system, for both brand-new DEPPACK and DEPPTINY. Seems to happen when
  saving last part. Let's debug it again. Ok, fixed too.

25-03-99    Thursday >
        Send everything... Some files were corrupted, and I'm told that
  current version is 0.22 instead of 0.21. Make the changes and resend it
  all.

26-03-99    Friday
        It seems that a really strange bug now happens only under DOS and
  not under Windows, corrupting last part of a big test file. This version
  seems doomed.

 6-04-99    Tuesday
        3 more bytes from DEPPACK has been removed. Tomorrow I'll send it

27-04-99    Tuesday >
        Send DEPPACK and DEPPTINY to aPACK mail-list to see if someone
  knows what is happening.

26-05-99  > Wednesday
        aPLib 0.22b has finally been released. As it happened with 0.20,
  some files aren't updated. Jibz didn't respond about the strange bug when
  I asked him information in the mail-list, and there's no reference about
  it in the release. Maybe it wasn't DEPPACK problem after all. Tomorrow
  I'll send him newer versions of updated files.

28-05-99  > Friday >
        I've finally sent the updated files, and a silent update with those
  files has been made in Jibz's homepage.

 1-06-99  > Tuesday
        Oleg Prokhorov has found the famous bug. Microsoft fault, because:
  wasn't DOS supposed to be a 16 bit O.S.? Then why the hell is it
  corrupting the high part of eax when calling an API function? The fix
  cost 2 bytes for each version, so it makes DEPPTINY grow to 513 (arg!).

 3-06-99    Thursday
        I think we should sacrifice again the cld intruction in DEPPTINY
  to keep it in 512

13-07-99    Tuesday
        I've found 2 small mistakes in the files I sent for the updated
  version. DEPACK16.NAS claims to be v0.21b, and DEPACK16.ASM has a
  "adc ecx, ecx" instead the "adc cx,cx" it should have. (So TASM users
  get a 1 byte bigger version than NASM ones). Removed 8 bytes from
  DEPPACK. Now it's 954/878 bytes long.

15-07-99    Thursday >
        Send everything

20-07-99  > Tuesday
        Jibz has managed to cut 11 bytes from DEPPTINY!!!! Of course cld
  is welcomed again, leaving it in 502.

25-07-99    Sunday
        Ported Jibz's last optimizations to DEPPACK (941/873) and source
  codes, and added back the missing bad infile test to DEPPTINY (506).
  Oops! I changed the wrong "adc ecx,ecx" to "adc cx,cx", fixed. Tomorrow
  I'll send it.

 2-09-99    Thursday
        Removed 1 byte from DEPPACK (940/872)

22-11-99    Monday
        3 bytes removed from DEPPTINY (503: Getting near the 500 barrier...)
  and DEPPACK (937/833 -> Using Chut's 4C, algorithm 02).

24-11-99    Thursday
        The new 4C-02 v1.01 leaves DEPPACK at 832.

28-11-99    Monday
        The updated 16 bit depackers finally appear at Jibz's homepage.

29-02-00    Tuesday
        Removed 2 more bytes from DEPPTINY (501) and DEPPACK (935/827 ->
  Using Chut's 4C-02 version 1.00 alpha). Start commenting DEPPTINY a bit
  more...

 2-03-00    Thrusday
        Removed another byte from both (500, 934/826).

 4-03-00    Saturday
        And one more (499!, 933/826). I'm thinking about performing a
  memory test to check you actually have those 162K needed to run the
  programs. I guess 13 bytes are more than enough for DEPPTINY.

10-03-00    Friday
        Removed 2 more bytes from both again (497,931/822).

 3-04-00    Monday
        Removed 2 VERY evident do-nothing bytes from DEPPACK (929/821).

 4-04-00    Tuesday >
        Added memory check (7 bytes for DEPPTINY (504), 29 for DEPPACK
  (962/848)). Continue adding comments to DEPPTINY... and put'em in DEPPACK
  too. Now I feel it's small, safe and clear enough to raise the version to
  v0.9 (nearly perfect?).

 5-04-00  > Wednesday
        Jibz cutted 2 bytes from core, and 6 more from DEPPACK's syntax
  text, leaving sizes of 502 for DEPPTINY and 954/841 for DEPPACK. Also,
  aPLib's version number has been increased to v0.26b.

25-04-00    Tuesday
        aPLib 0.26 has been released.

 8-05-00    Monday >
        1 byte removed from core. One bug detected in TASM version of
  DEPACK16.ASM

14-05-00    Sunday
        Removed another byte from DEPPTINY (500).

23-05-00    Tuesday >
        Removed another byte from both DEPPTINY (499) and DEPPACK (952/836).
  My mail server has changed from bart to aluesi, so my email is now:

  [e-mail]

 2-07-00    Sunday
        Removed another byte from both DEPPTINY (498) and DEPPACK (951/836)

 3-07-00    Monday >
        Another one! (497,950/831)

 8-07-00    Saturday
        Another one! (496,949/831)

14-10-00    Saturday
        BUGFIX: Up till now, the depackers didn't work with a drive for
  outfile. I've been able to fix this, and costed 17 bytes for DEPPTINY
  (513, aargh! CLD sacrifice again?). aPPACK also suffers from this bug.

15-10-00    Sunday
        No sacrifice. I removed one more byte -> 512!. Applied bugfix to
  DEPPACK (967/858). Version raised to v0.99 (now, I will just add timestamp
  preservation for v1.00, only for DEPPACK, as DEPPTINY can't handle any
  more code).

19-10-00    Thrusday >
        Added timestamp preservation to DEPPACK (983/864). Version raised
  to v1.0 . Next improvement: long filenames support?

27-10-00    Friday >
        Removed 9 bytes from DEPPACK (974/851), and 7 from DEPPTINY (505).

29-10-00    Sunday >
        Removed 4 more bytes from DEPPTINY (501) and 7 from DEPPACK
  (967/851 (Hmmm, no gain after compression). Now the timestamp preservation
  nearly fits in DEPPTINY!... 3 bytes left to be removed to add it.

23-02-01  > Friday
        Jibz sent me a beta version of new aPLib 0.34, and asked me to
  update my depackers, because the encoding method has changed, and also
  the aPPack program has added a header to the files. Well, it will take
  just a few bytes to update the decode routine, so it would fit in
  DEPPTINY, but the header adds even more bytes, so I feel I can't fit it
  under 512, so I'll discontinue development of DEPPTINY and concentrate
  just on DEPPACK. But perhaps a headerless DEPPTINY will be included for
  anyone who wants to see the last optimizations.

25-02-01    Sunday
        Updated the decoding routine. Just 10 bytes. It works cutting the
  first 8 bytes from new packed files. DEPPTINY is ready.

28-02-01    Wednesday >
        Added header support, of course including more tests for bad infile
  detection based on header. It makes 1023/903 for DEPPACK. Version raised
  to 1.1 . Version 1.2 will probably include partial long filenames support.

19-10-01  > Friday
	Received a new beta of aPLib 0.34. Header size has grown from 8 to 
  24 bytes, adding CRC info and info about packed file.

22-10-01    Monday >
	Made a few changes to DEPPACK to support new headers, but new tests
  using CRCs and size of packed file are not added... yet!. No size penalty.

24-10-01    Wednesday >
        DEPPACK now reads the header in two steps, to skip the header
  whatever it's size is, making it compatible with future aPPack versions.
  This costs a few bytes (1034/913).