Technotools/Technotools (Chestnut CD-ROM)(1993).ISO/unix/unzup41e/history.41

Text File | 1991-05-13 | 15.7 KB | 398 lines

unzip version 4.1, 13 May 91

- Cave Newt, Info-ZIP #139
These are the minor things I found:
   (1) the do-nothing file_io.c umask patch got included by mistake
   (2) those same two lines in unzip.c are shifted over again (somebody's
        patch program REALLY doesn't like those lines...they're full of
        feelthy tabs, which may have something to do with it)
   (3) the man page wasn't quite updated to reflect the -z and -o options
   (4) the Makefile now has two msc_os2 targets

I've fixed (1) through (3) in this patch.  None of them actually has any
impact on the code--(2) is obviously cosmetic, and (1) is effectively so,
although it's potentially dangerous.  For the man page I just updated the
nroff source (unzip.1); unzip.man can be regenerated.  In addition to in-
cluding "o" and "z" in the synopsis and adding the -z description, I also
figured I might as well put in the "4.1" comments, so I did that.
[Affected file_io.c, unzip.c, unzip.1]

- Jim Dumser, Info-ZIP #140, 7 May 91
First is my "proposed" umask patch, after which I conceded to Greg's
fix.  It was still included, and altho it doesn't hurt anybody, it
really shouldn't be in there.  [NOP, fixed by Cave Newt's patch above.]

Second is something that will hurt people.  There were 2 msc_os2 targets
in the Makefile.  I picked the one that was submitted last (and for
which the .def file was included), and removed the other.
[Affected Makefile]

-Cave Newt, 8 May 91 (private message, not in Info-ZIP)
Yes, it's YAVU (Yet Another VMS Update).  I edited the VMSNOTES file
a little bit to reflect what I've lately figured out about the "variable
length, 510 byte max" format, and I undid the interior vmsattr.zip file
since there wasn't much need for it as a separate sub-sub-archive.  No
big changes, but we may as well get ready for the release.
[New VMS.zip]

- Cave Newt, 10 May 91 (private message, not in Info-ZIP)
First of all, one last (?) change to the man page:  I forgot to update
the info for the -a switch, and it still contained your old e-mail
address.  Good thing you ignored my impatient nagging, huh? :-)  Patch
below to the patch I already sent you...
[Affected unzip.1]

And from the Coordinator:
We're going public with this one, folks.  History, here we come!

================

v4.07 BETA, 4 May 91

- (That's "05-04-91" in feelthy civilian format)

- Starting with v4.06 source, patched in the v4.06a stuff
  (see UNZ_406A.ZIP patch file)

- Wim Bonner's OS2 Makefile patch (for long file names)
  and an "unzip.def" file.  (MAKEFILE.127)
- Cave Newt: fix for Makefile's pyramid target (MAKEFILE.406)
- Mark Adler's minor diffs to make it compile cleaner and smaller
  on the NeXT (plus trimmed out some feelthy tabs).  (UNZIP.125)

- The Great Umask Wars commence, with Bo Kullmar, Cave Newt,
  and Jim Dumser all contributing to the Infamous Info-ZIP MindFog.
  (It all worked out in the end .. kinda ..)
  UNZIP_C.128: Bo's first shots at Fort Sumpter; UNZIP_C.129: Cave
  Newt throws in a few curves.

- Bo Kullmar adds a -z option to only display the archive comment.
  He also notices the Infamous 8192 bug (an extra LF is inserted
  after every 8192 byte in the do_string function.
  (UNZIP_C.132)

- Info-ZIP Coordinator prepares lengthy curses for anyone preparing
  to write 8192-byte-long archive comments.  (Private files)

- Jim Dumser then leaps into the -z conflict.  He also discovers
  who's sneaking in feelthy tabs to our source!
  He then coins the famous phrase:
  "needs to be applied either before Bo's patch ("A -z option to unzip",
  91.129) or after my patch ("Re:  A -z option to unzip", 91.130).
  Actually, because the 91.129 patch does more than just add the COMMENT
  option, you really need to apply both Bo's and mine, then apply this."
  (Which *almost* invokes Info-ZIP Coordinator's curses.)
  (UNZIP_C.133, UNZIP_C.134)

- Jim Dumser, returning to the Umask Front, discovers there *ain't*
  no feelthy umask() in Turbo C!  And MSC fakes it.  umask()-related
  code is #ifdef'ed for Unix systems only.  (DUMSER.407)
- Cave Newt chucks a grenade into umask() with yet more #ifdefs.
  (407.MSG)
- Jim Dumser pretends to clear up the confusion.  Contributes a stupendous
  diff or two just to be sure I got it all.  Bo Kullmar throws in one
  of them too.  Looks like I got it all.

================

v4.06a, late April 91

 - unzip Makefile:  fix for pyramid target from Cave Newt.
   Somebody switched "make" with "$MAKE" or some such.

- Makefile Change for OS/2 (and addition of def file)
  (Wim Bonner).  Tweak to the msc_os2 system of the Makefile,
  plus a wee little def file for OS/2.

- Unzip 4.06 diffs for the NeXT (Mark Adler)
  "Here are some very minor diffs for Unzip 4.06 to make it compile cleaner
  and smaller on the NeXT.  (By the way, there were feelthy tabs in unzip.c,
  unzip.h, and a few other non-source files.)"

================

4.06, 25 April 1991

- Minor patches to Makefile (Jim Dumser, Info-ZIP #096
- And then a BIG change to Makefile: a "generic" Makefile
  From Cave Newt (Info-ZIP #099)
- Case-conversion-info patch from Cave Newt (Info-ZIP #098):
  "Unzip does 'smart' lowercasing of filenames in zipfiles,
  a feature I like a lot (that's why I put it in :-).  But zip
  doesn't do any conversions, and it's easy to forget sometimes
  that all the names in a zipfile are really uppercase.  This is
  a PROPOSED patch that lets you know when case-conversion has
  occurred in unzip."
  Well, the Newt proposed .. and nobody screamed .. so it becomes
  gospel.
- VMS file attribute patch (Cave Newt, Info-ZIP #098):
  Wee little patches to the unix code, and some new files in the
  VMS.zip file.
- Wim Bonner, Info-ZIP #100, reported a "missing semicolon" error
  with MSC 6.0 and OS/2:
  "The problem is that there are a set of local variables declared in
  conditional defines, and they are after some code has already appeared
  in the subroutine.
  "If the conditional block is moved before the small amount of code, the
  file compiles OK."
  Fixed.
- Jim Dumser (Info-ZIP #???, 26 Mar), further patching Cave Newt's
  file attributes patch with more of his own.
- Info-ZIP #106:  Patched to solve problem with truncated comments.
- Added "-o" and "-n" patches by Gene McManus (Info-ZIP #106),
  to permit overwriting output files without prompting.
  These switches do virtually the same thing (so I don't know why
  he did them both) .. and the patches kinda conflict .. but I made
  it fit.
- Mac diffs from Johnny Lee (Info-ZIP #115):
  "Catching up with the permissions changes. Since I don't have any
  AppleShare docs handy, this patch only determines if a file
  is read-only or not."
- Cave Newt (Info-ZIP #120) fixes the -a switch usage screen.
- Cave Newt (Info-ZIP #121) *really* fixes the truncated zip comment
  (superseding my fix in #106).

Archive still includes the MAC, VMS, and ATARI-specific code in
their own separate .ZIPs.

Last patches before public v4.10 release (I hope)

David Kirschbaum
Info-ZIP Coordinator

================

4.05, 19 Mar 91

[We're going to use Greg Roelof's version, since his is probably the
most complete.  From the header of his message: ]

Here's the package, in two parts.  The zipfile is basically complete,
except that it doesn't have a History.405 file or an unzip405.descr
file.  I've grouped the non-Unix-OS stuff into four sub-zipfiles inside
the main one:  one each for Atari ST, Mac, MS-DOS, and VMS.  This keeps
things a little cleaner and prevents collisions between the Atari Turbo
makefiles and the MS-DOS ones.  You can take them out for the beta
version if you want, but they aren't really all that big.  Note that
the Mac and Atari binaries are NOT included; the little zipfiles only
contain extra source files, notes, and makefile stuff.

Now, a complete list of the changes to 4.05 from 4.04:

  Missing 4.04 patches:
    - Antoine's FILENAME_MAX fix
        - size-of-encrypted-member (negative size factors) fix
    - Bo's missing <sys/timeb.h> for BSD moved into unzip.h

  Bug fixes:
    - setting the time on stdout
    - listing of three digits of 21st century years instead of two
    - multiple usage screens for non-existent switches (e.g., unzip -XY)
    - [VMS] inability to set file dates when extracting directory trees
    - [VMS] choke-and-die on multiple-dot filenames (now converted to '_')

  New features:
    - BIG Mac attack from Johnny Lee
    - (correct) handling of file permissions, both Unix zip and PKZIP
    - new -V option to retain version numbers from VMS-created zipfiles
    - ifdef'd code to deal with PAK bug ("on" for now)
    - detection of big-endian and/or struct-padding behavior, and noti-
       fication to user to recompile if necessary

  New Makefile targets:
    - msc_os2 from Wim Bonner
    - sequent from Phil Howard
    - ultrix split into ultrix_vax and ultrix_risc
    - dec5820 commented out (ultrix_risc system)
    - renames:  diab -> dnix; cray_scc -> cray


NOT included in this release are the Atari patches; partly this is because
the majority of them had nothing to do with Atari at all and should have
been posted to the group, and partly it is because the remaining ones, which
DID have to do with Atari, largely consisted of the removal of all the non-
Atari operating systems (like Unix, DOS, VMS; Mac didn't exist at the time).

Also not included are (1) the proposed "generic"/"generic2" makefile targets
(still waiting for some Info-ZIP feedback); (2) the ultimate VMS file-type
fix on which I've been working (close, still waiting on someone); and (3) a
new case-conversion-warning patch I'll post shortly to see what folks think.
(Oh, and the two 4.05 description files I mentioned above.)

I don't *think* I missed any unzip patches that have been posted, but people
should check carefully to see if their favorite one is in here.  Also, the
Mac patches collided rather violently with my own, so folks should keep their
eyes peeled.  I tried to hand-tweak everything, but patch may have snuck one
or two into a slightly wrong place.  No crashes yet on MY system...

Finally, with the exception of the PAK-fix, I think I've removed everything
that hasn't been posted in one way or another (i.e., my own little tweaks).
So this should be a nice, clean release, which should in turn make everyone
very happy.

Greg

================

v4.04, 22 Feb 91

- Makefile:
  From: Michael Regoli <[e-mail]>
  It looks as if "make" is not defined for the IBM RT.
  We need a MAKE macro definition on or about line 67:
  (And he sent a context diff.  Applied.)

  From: [e-mail] (Bruce Kahn)
  I just got a copy of 4.03 from the archives and it seems that my little
  addition to the Makefile from 3.1x to support Data General AViiONs was
  lost.  Im enclosing a diff -c on the 'current' 4.03 makefile to add this
  back in.
  (Applied after slight editing.)

  And someone said Suns needed "-DBSD" on their make line (maybe
  Josh? Bo Kullmar?).  Anyway, Suns are now broken out onto their own line
  (basically "_16", plus "-DBSD").

- file_io.c (The date/time problem, fixed by Bo Kullmar):
  Lots of changes (confusing), so you'd want the original message
  and patch if you want details.  Usually named file_io.404.
  Bo's patch forgot one thing: a reference to /sys/timeb.h for BSD systems.
  (diff included and applied)

- New VMS files from Cave Newt (entirely replacing old ones, and one
  VMSSHARE.OPT or some such deleted entirely).

- unzip.c

  From: Cave Newt <[e-mail]>
  Subject: Compression ratio calculation bug

  Ah, yes, the old negative-ratio bug.  Below is a patch to fix this;
  here's the explanation:
  [explanation deleted, patch applied.]

================


v4.03, 8 Dec 90

- Changes to VMSNOTES.
- Patched via Newt's "even more stupendous(ly long) 4.03 patches"
- Entirely new unzip.c and unzip.h
- misc.c, mapname.c and file_io.c patched via patch.403
- Fixes are for VMS: reversion to block-boundary seeks while
  retaining the random-access behavior that gives (n)unzip its speed.
- Routines to find end-of-central-directory sig rewritten and
  streamlined into a single function.
- New LSEEK macro added.
[Full details for all this horridness are in the unz403_p patch archive.]

- Fix involving making the filename longer
- fixing a leap-year bug (2000 IS a leap year).
- Consolidated more repeated error strings (unzip.c)
- Removed an unused macro (UPDCRC32).

================

v4.02, 6 Dec 90

- Per Cave Newt's request, changed nunzip's in VMS_MAKE.COM
  to unzip.

- mapname.c :  History of changes removed.

- Many changes, patches in Cave Newt's big unzip_c.401, unz402_p.zip,
  to include more extensive explanation and comments.

- Bug in unzip.c related to encryption:
  Mark and Newt were getting "bus errors" (sic) when unzip tried to test
  or unzip an encrypted file. (Probably array overruns)
  Solution:  check general purpose bit flag to see if file is encrypted.
  If so, skip it (with regretful message).
  Later we may decrypt (Mark offers his routine).

- -t option change:
  The -t option in 4.01 currently prints both the individual file "OKs"
  and a summary; this is redundant.  So now -t just prints the "OKs"
  (like the old versions), -tq just prints the summary, and -tqq
  doesn't print anything at all.

- Misc indentation corrected.

- man page changed:
  Separated the modifier options into their own little heading.
  Expanded the -a explanation to include EBCDIC, since that's why it was
  originally done in the first place.
  Added 3.1 to the list at the bottom.

- Minor centralization of repeated error message strings.

================

v4.01, 4 Dec 90:

1 - From Down Under:
From: [e-mail]

The makefile.msc as it stands more or less works for TurboC 2.0,
so makefile.tc is probably redundant.  However, TurboC throws what
we colloquially call a `wobbly' over one line in file_io.c: where
`>> (space) =' is used instead of the ANSI `>>=' construct.  The
fix is easy (appended).

[Makefile.msc patched to include both Turbo C and MSC compiles,
 renamed to MAKEFILE.DOS.  Actually, it's for OS/2 also.
 Works just fine for Turbo C v2.0, can't test the others.
 Removed MAKEFILE.TC.
]

And the file_io.c patch was a one-liner at line 312 /318:
!     bitbuf >> = bits;
changed to:
!     bitbuf >>= bits;

================

2 - From: Cave Newt <[e-mail]>
    (details in unzip400.msg, unz401_p.zip)

- a patch to 3.99 which you missed and I forgot about:  the name[STRSIZ]
  one in mapname.c.
  [Extracted to mapname.pch, mapname.c patched.]

- indent11 (AKA "Mr. Mangler") likes tabs, which is going to play hell
  with patches.

  [I'm using indent11 with the "-kr" settings.
   And all source will be archived with NO tabs.
   Except for the stupid makefiles, however
  ]

- "makefile.msc" should have been uppercase.
  [It's MAKEFILE.DOS now because of TC/MSC/OS2 integration.]

- man page still says "3.1" at the top and has a weird (old)
  e-mail address for you at the bottom.
  [unzip.1 edited to reflect "4.0".]

- Comments at the top of unzip.c say "unzip40.tar.Z" rather than
  "unzip400".
  [Fixed]

================
Final 4.01 Toad Hall Note:
- Compiling in Turbo C integrated environment (e.g., using the .PRJ
  file) produced an UNZIP.EXE a couple hundred bytes smaller than the
  one compiled via tcc and the MAKEFILE.DOS.  Obviously I've forgotten
  a switch somewhere, or my default tcc settings are slightly off.
  So if you want the absolute smallest, suggest integrated compile.
- No more stupid mixing of Unix and DOS text files.  EVERYTHING's gonna
  have (1) no tabs and (2) Unix LF line endings.  If you're gonna work
  in DOS, you'd better have FLIP or TOADCRLF or some such to convert
  these Unix files.
  We'll still indicate DOS files vs. Unix files via name case (e.g.,
  DOS files'll be all uppercase).  Of course, the VMS files are ALSO
  uppercase (meant to ask you about that, Cave Newt), but that's ok.
  For the DENSA members, anything starting with VMS is NOT DOS, ok?
================