oweals/openssl.git
25 years agoNew UTF8 utility functions to parse/generate UTF8 strings.
Dr. Stephen Henson [Sat, 4 Sep 1999 17:19:55 +0000 (17:19 +0000)]
New UTF8 utility functions to parse/generate UTF8 strings.

25 years agoReinitialize global variables when necessary (for monolith application).
Bodo Möller [Fri, 3 Sep 1999 23:08:45 +0000 (23:08 +0000)]
Reinitialize global variables when necessary (for monolith application).

25 years agouse explicit constant 11 just once
Bodo Möller [Fri, 3 Sep 1999 22:37:38 +0000 (22:37 +0000)]
use explicit constant 11 just once

25 years agoMake previous bugfix actually work
Bodo Möller [Fri, 3 Sep 1999 16:49:11 +0000 (16:49 +0000)]
Make previous bugfix actually work

25 years agoFix server behaviour when facing backwards-compatible client hellos.
Bodo Möller [Fri, 3 Sep 1999 16:33:11 +0000 (16:33 +0000)]
Fix server behaviour when facing backwards-compatible client hellos.

25 years ago-no_dhe option for ssltest.c
Bodo Möller [Fri, 3 Sep 1999 16:31:36 +0000 (16:31 +0000)]
-no_dhe option for ssltest.c

25 years agoUse closesocket macro consistently, not close directly, for easier
Bodo Möller [Fri, 3 Sep 1999 14:06:09 +0000 (14:06 +0000)]
Use closesocket macro consistently, not close directly, for easier
portability.
Submitted by: Lennart Bång

25 years agoHandle "#if 0" correctly (I hope)
Bodo Möller [Fri, 3 Sep 1999 13:30:47 +0000 (13:30 +0000)]
Handle "#if 0" correctly (I hope)

25 years agoAdd new 'spkac' utility and several SPKAC utility functions.
Dr. Stephen Henson [Fri, 3 Sep 1999 01:08:34 +0000 (01:08 +0000)]
Add new 'spkac' utility and several SPKAC utility functions.

25 years agoMake DH_free() free up any ex_data and also call the finish method.
Dr. Stephen Henson [Wed, 1 Sep 1999 23:50:43 +0000 (23:50 +0000)]
Make DH_free() free up any ex_data and also call the finish method.

25 years agoRIPEMD160 shape-up. Final touch.
Andy Polyakov [Sat, 28 Aug 1999 13:18:25 +0000 (13:18 +0000)]
RIPEMD160 shape-up. Final touch.

25 years agoRIPEMD160 shape-up Intel assembler companion. Cycle counter benchmarks
Andy Polyakov [Sat, 28 Aug 1999 13:07:51 +0000 (13:07 +0000)]
RIPEMD160 shape-up Intel assembler companion. Cycle counter benchmarks
went down from 1050 to 921 cycles on Pentium II. I haven't checked the
figures on Pentium yet.

25 years agoRIPEMD160 shape-up. Major news are that it's operational on all platforms
Andy Polyakov [Sat, 28 Aug 1999 12:55:45 +0000 (12:55 +0000)]
RIPEMD160 shape-up. Major news are that it's operational on all platforms
now and I'm putting it back to 'make test' later today.

25 years agomake testapps after the library.
Ulf Möller [Sat, 28 Aug 1999 12:50:48 +0000 (12:50 +0000)]
make testapps after the library.

25 years agomd32_common.h update and accompanying MD5 update.
Andy Polyakov [Sat, 28 Aug 1999 12:41:03 +0000 (12:41 +0000)]
md32_common.h update and accompanying MD5 update.

25 years agoAllow the extension section specified in config files to be overridden
Dr. Stephen Henson [Fri, 27 Aug 1999 00:08:17 +0000 (00:08 +0000)]
Allow the extension section specified in config files to be overridden
on the command line for various utilities.

25 years agoAllow the 1.OU="my OU" syntax in 'ca' for SPKACs.
Dr. Stephen Henson [Wed, 25 Aug 1999 23:18:23 +0000 (23:18 +0000)]
Allow the 1.OU="my OU" syntax in 'ca' for SPKACs.

25 years agoAllow extensions to be added to certificate requests, update the sample
Dr. Stephen Henson [Wed, 25 Aug 1999 16:59:26 +0000 (16:59 +0000)]
Allow extensions to be added to certificate requests, update the sample
config file (change RAW to DER).

25 years agomake update.
Ulf Möller [Tue, 24 Aug 1999 17:45:16 +0000 (17:45 +0000)]
make update.

25 years agoMinor MIPS III/IV tune-up.
Andy Polyakov [Tue, 24 Aug 1999 16:02:16 +0000 (16:02 +0000)]
Minor MIPS III/IV tune-up.

25 years agoFix for a bug which meant encrypting BIOs sometimes wouldn't read the final
Dr. Stephen Henson [Tue, 24 Aug 1999 13:21:35 +0000 (13:21 +0000)]
Fix for a bug which meant encrypting BIOs sometimes wouldn't read the final
block.

25 years agoInitial support for DH_METHOD. Also added a DH lock. A few changes made to
Dr. Stephen Henson [Mon, 23 Aug 1999 23:11:32 +0000 (23:11 +0000)]
Initial support for DH_METHOD. Also added a DH lock. A few changes made to
DSA_METHOD to make it more consistent with RSA_METHOD.

25 years agoInitial support for DSA_METHOD...
Dr. Stephen Henson [Sun, 22 Aug 1999 17:57:38 +0000 (17:57 +0000)]
Initial support for DSA_METHOD...

25 years agoAllow memory bios to be read only and change PKCS#7 routines to use them.
Dr. Stephen Henson [Thu, 19 Aug 1999 13:07:43 +0000 (13:07 +0000)]
Allow memory bios to be read only and change PKCS#7 routines to use them.

25 years agoFix horrible (and hard to track down) bug in ssl23_get_client_hello:
Bodo Möller [Wed, 18 Aug 1999 17:14:42 +0000 (17:14 +0000)]
Fix horrible (and hard to track down) bug in ssl23_get_client_hello:
In case of a restart, v[0] and v[1] were incorrectly initialised.
This was interpreted by ssl3_get_client_key_exchange as an RSA decryption
failure (don't ask me why) and caused it to create a _random_ master key
instead (even weirder), which obviously led to incorrect input to
ssl3_generate_master_secret and thus caused "block cipher pad is
wrong" error messages from ssl3_enc for the client's Finished message.
Arrgh.

25 years agoReally undo the base64 change so that make test survives
Bodo Möller [Wed, 18 Aug 1999 16:42:27 +0000 (16:42 +0000)]
Really undo the base64 change so that make test survives

25 years agoFix PKCS7_ENC_CONTENT_new() to include a sensible default content type and add
Dr. Stephen Henson [Tue, 17 Aug 1999 12:58:01 +0000 (12:58 +0000)]
Fix PKCS7_ENC_CONTENT_new() to include a sensible default content type and add
support for encrypted content type in PKCS7_set_content().

25 years agoReturn 0 for succesful exit when -noout is used.
Bodo Möller [Tue, 17 Aug 1999 08:33:59 +0000 (08:33 +0000)]
Return 0 for succesful exit when -noout is used.

25 years agoUndo base64 decoding change (was not a bug fix).
Ulf Möller [Sat, 14 Aug 1999 20:49:37 +0000 (20:49 +0000)]
Undo base64 decoding change (was not a bug fix).

25 years agoediting error.
Ulf Möller [Fri, 13 Aug 1999 21:02:25 +0000 (21:02 +0000)]
editing error.

25 years agoHonor $PERL environment variable in Configure.
Ulf Möller [Fri, 13 Aug 1999 20:48:32 +0000 (20:48 +0000)]
Honor $PERL environment variable in Configure.

Submitted by: Even Holen <Even.Holen@itea.ntnu.no>

25 years agoFix faulty base64 decoding of data that was 46 or 47 bytes long.
Ulf Möller [Fri, 13 Aug 1999 19:42:33 +0000 (19:42 +0000)]
Fix faulty base64 decoding of data that was 46 or 47 bytes long.

Submitted by: Ivan Nejgebauer <ian@uns.ns.ac.yu>

25 years agoAdd pkcs7 and des apps to "make all".
Ulf Möller [Fri, 13 Aug 1999 19:21:53 +0000 (19:21 +0000)]
Add pkcs7 and des apps to "make all".

25 years agoCompile pkcs7 and des apps.
Ulf Möller [Fri, 13 Aug 1999 18:04:04 +0000 (18:04 +0000)]
Compile pkcs7 and des apps.

25 years agoDisable the text about foo.h => openssl/foo.h.
Bodo Möller [Thu, 12 Aug 1999 11:31:01 +0000 (11:31 +0000)]
Disable the text about foo.h => openssl/foo.h.
Everyone should have got it by now.

25 years agoRemove -DWINDOWS in debug configuration.
Bodo Möller [Thu, 12 Aug 1999 10:03:05 +0000 (10:03 +0000)]
Remove -DWINDOWS in debug configuration.

Submitted by: Erik Johannisson <erik@johannisson.se>

25 years agoSubmitted by: Lidong Zhou <ldzhou@cs.cornell.edu>
Bodo Möller [Wed, 11 Aug 1999 13:31:03 +0000 (13:31 +0000)]
Submitted by: Lidong Zhou <ldzhou@cs.cornell.edu>

25 years agoAdd functions to allow extensions to be added to certificate requests.
Dr. Stephen Henson [Wed, 11 Aug 1999 13:08:58 +0000 (13:08 +0000)]
Add functions to allow extensions to be added to certificate requests.

Modify obj_dat.pl to take its files from the command line. Usage is now
perl obj_dat.pl objects.h obj_dat.h
this should avoid redirection shell escape problems under Win32.

25 years agoFix typo.
Bodo Möller [Wed, 11 Aug 1999 08:28:40 +0000 (08:28 +0000)]
Fix typo.

25 years agoUpdates.
Bodo Möller [Tue, 10 Aug 1999 11:36:22 +0000 (11:36 +0000)]
Updates.
Prototypes and constant declarations for non-copying reads and writes for
BIO pairs (which is totally untested as of now, so I don't yet commit
the actual source code, but reserve the numbers to avoid conflicts).

25 years agoSupport for parsing of certificate extensions in PKCS#10 requests: these are
Dr. Stephen Henson [Mon, 9 Aug 1999 22:38:05 +0000 (22:38 +0000)]
Support for parsing of certificate extensions in PKCS#10 requests: these are
used by things like Xenroll. Also include documentation for extendedKeyUsage
extension.

25 years agoBIO_write and BIO_read could, in theory, return -2.
Bodo Möller [Mon, 9 Aug 1999 16:33:34 +0000 (16:33 +0000)]
BIO_write and BIO_read could, in theory, return -2.

25 years ago-crlf option.
Bodo Möller [Mon, 9 Aug 1999 13:01:48 +0000 (13:01 +0000)]
-crlf option.

25 years agoProvide fixed seed for parameter generation to speed up -dhe1024.
Bodo Möller [Mon, 9 Aug 1999 12:59:10 +0000 (12:59 +0000)]
Provide fixed seed for parameter generation to speed up -dhe1024.

25 years agoBump after tarball rolling.
Ralf S. Engelschall [Mon, 9 Aug 1999 11:14:08 +0000 (11:14 +0000)]
Bump after tarball rolling.
Friends, feel free to start again hacking for 0.9.5... ;)

25 years agoBump version to 0.9.4 OpenSSL_0_9_4
Ralf S. Engelschall [Mon, 9 Aug 1999 10:40:38 +0000 (10:40 +0000)]
Bump version to 0.9.4

25 years agoOk, I'll do a few more final platform tests and then I'll roll the 0.9.4
Ralf S. Engelschall [Mon, 9 Aug 1999 10:25:36 +0000 (10:25 +0000)]
Ok, I'll do a few more final platform tests and then I'll roll the 0.9.4
tarball baby for us. PLEASE DO NOT COMMIT ANYTHING UNTIL YOU SEE MY FINAL
COMMITS TO `STATUS' INDICATING THAT WE'VE REACHED 0.9.5-dev.

25 years agoOk, final fix for `config' script to detect all flavors of FreeBSD
Ralf S. Engelschall [Mon, 9 Aug 1999 10:16:51 +0000 (10:16 +0000)]
Ok, final fix for `config' script to detect all flavors of FreeBSD
in a more general way.

25 years agospelling
Bodo Möller [Sun, 8 Aug 1999 22:41:24 +0000 (22:41 +0000)]
spelling

25 years agoInstall libRSAglue.a when OpenSSL is build with RSAref.
Ralf S. Engelschall [Sun, 8 Aug 1999 19:12:26 +0000 (19:12 +0000)]
Install libRSAglue.a when OpenSSL is build with RSAref.
This should now finally make the RSAref users happy...

25 years agoComments.
Bodo Möller [Sun, 8 Aug 1999 14:21:04 +0000 (14:21 +0000)]
Comments.

25 years agoadd comments
Bodo Möller [Sun, 8 Aug 1999 14:07:30 +0000 (14:07 +0000)]
add comments

25 years agomore consistent formatting
Bodo Möller [Sun, 8 Aug 1999 14:06:29 +0000 (14:06 +0000)]
more consistent formatting

25 years ago*** empty log message ***
Dr. Stephen Henson [Sun, 8 Aug 1999 12:51:19 +0000 (12:51 +0000)]
*** empty log message ***

25 years agoBackout the following change:
Ralf S. Engelschall [Sun, 8 Aug 1999 11:56:29 +0000 (11:56 +0000)]
Backout the following change:

| revision 1.109
| date: 1999/06/24 19:39:23;  author: ulf;  state: Exp;  lines: +0 -1
| Use the same CPU recogition method for FreeBSD 3 as for the other FreeBSD
| versions. (The FreeBSD and FreeBSD-elf Configure entries were identical.)

Because Ulf overlooked the ELF vs. a.out difference, i.e. the two entries were
_NOT_ identical! This way OpenSSL 0.9.4 at least compiles again under FreeBSD
2 and 3. For 0.9.5-dev I'll try to fix the stuff in a more general way to also
allow FreeBSD 4 and other variants.

25 years ago0.9.4
Bodo Möller [Sun, 8 Aug 1999 11:45:56 +0000 (11:45 +0000)]
0.9.4

25 years agoFix two remaining prototype-related warnings
Ralf S. Engelschall [Sun, 8 Aug 1999 11:25:32 +0000 (11:25 +0000)]
Fix two remaining prototype-related warnings

25 years agoRemove the perl stuff also optically from the build procedure,
Ralf S. Engelschall [Sun, 8 Aug 1999 10:29:52 +0000 (10:29 +0000)]
Remove the perl stuff also optically from the build procedure,
because it could confuse people (which then send us bug reports)...

25 years agoLet us prepare for tomorrow...
Ralf S. Engelschall [Sun, 8 Aug 1999 10:25:50 +0000 (10:25 +0000)]
Let us prepare for tomorrow...

25 years agoI don't think this is a real showstopper. Our internal verify procedure lacks
Ralf S. Engelschall [Sun, 8 Aug 1999 10:19:48 +0000 (10:19 +0000)]
I don't think this is a real showstopper.  Our internal verify procedure lacks
even more and I consider this more a useful feature request than a release
showstopper.

25 years agoA few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
Ralf S. Engelschall [Sun, 8 Aug 1999 10:15:43 +0000 (10:15 +0000)]
A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
Hint from: Andrija Antonijevic <TheAntony2@bigfoot.com>

25 years agoconsistent style
Ralf S. Engelschall [Sun, 8 Aug 1999 09:39:43 +0000 (09:39 +0000)]
consistent style

25 years agoNew option "-crlf" to s_client and s_server which tells them to convert
Bodo Möller [Sat, 7 Aug 1999 02:51:10 +0000 (02:51 +0000)]
New option "-crlf" to s_client and s_server which tells them to convert
LFs into CRLFs when forwarding data from stdin to the TLS connection.
This is necessary for properly talking HTTP.
Because of the code freeze this change is by default disabled for now;
without -DAPPS_CRLF, the code is exactly as before.

25 years agoadd some more entries for 0.9.4
Bodo Möller [Sat, 7 Aug 1999 02:46:49 +0000 (02:46 +0000)]
add some more entries for 0.9.4

25 years agochange formatting a bit
Bodo Möller [Sat, 7 Aug 1999 02:45:31 +0000 (02:45 +0000)]
change formatting a bit

25 years agoFix -startdate and -enddate arguments to 'ca' program. Also update NEWS file
Dr. Stephen Henson [Fri, 6 Aug 1999 21:47:09 +0000 (21:47 +0000)]
Fix -startdate and -enddate arguments to 'ca' program. Also update NEWS file
with some 0.9.4 changes.

25 years agofix the bug
Bodo Möller [Fri, 6 Aug 1999 12:27:39 +0000 (12:27 +0000)]
fix the bug

25 years agoDont' assume that something starting with '-' is a filename --
Bodo Möller [Fri, 6 Aug 1999 11:18:44 +0000 (11:18 +0000)]
Dont' assume that something starting with '-' is a filename --
"openssl gendsa -help" now prints the usage summary, not error
messages that now file -help was found.

25 years agotypo while I poke around...
Ralf S. Engelschall [Thu, 5 Aug 1999 13:31:42 +0000 (13:31 +0000)]
typo while I poke around...

25 years agoDSA_dup_DH.
Bodo Möller [Thu, 5 Aug 1999 11:53:07 +0000 (11:53 +0000)]
DSA_dup_DH.

25 years agoNew function DSA_dup_DH, and fixes for bugs that were found
Bodo Möller [Thu, 5 Aug 1999 11:50:18 +0000 (11:50 +0000)]
New function DSA_dup_DH, and fixes for bugs that were found
while implementing and using it.

25 years agoAdd a wish.
Bodo Möller [Thu, 5 Aug 1999 00:15:15 +0000 (00:15 +0000)]
Add a wish.

25 years agoChange all the -'s to _'s.
Dr. Stephen Henson [Thu, 5 Aug 1999 00:11:05 +0000 (00:11 +0000)]
Change all the -'s to _'s.

25 years agoConfigure fix for Windows: under Windows+perl "system" ends up going via the
Dr. Stephen Henson [Wed, 4 Aug 1999 22:52:46 +0000 (22:52 +0000)]
Configure fix for Windows: under Windows+perl "system" ends up going via the
command shell which means that redirecttion needs backslashes in the paths.

25 years agoimprove clarity of instructions
Bodo Möller [Tue, 3 Aug 1999 20:19:04 +0000 (20:19 +0000)]
improve clarity of instructions

25 years ago0.9.4 won't be completed in July ...
Bodo Möller [Tue, 3 Aug 1999 12:24:14 +0000 (12:24 +0000)]
0.9.4 won't be completed in July ...

25 years agoMore diagnostics.
Ben Laurie [Tue, 3 Aug 1999 10:19:02 +0000 (10:19 +0000)]
More diagnostics.

25 years agoMake it compile under -pedantic.
Ben Laurie [Tue, 3 Aug 1999 10:18:27 +0000 (10:18 +0000)]
Make it compile under -pedantic.

25 years agocomment
Bodo Möller [Tue, 3 Aug 1999 08:31:13 +0000 (08:31 +0000)]
comment

25 years agoRevert erroneous change.
Bodo Möller [Tue, 3 Aug 1999 06:05:54 +0000 (06:05 +0000)]
Revert erroneous change.

25 years agoautomatically use no-mdc2 if no-des is requested.
Bodo Möller [Mon, 2 Aug 1999 22:13:46 +0000 (22:13 +0000)]
automatically use no-mdc2 if no-des is requested.

25 years agoRevert previous change -- it was an accident.
Bodo Möller [Mon, 2 Aug 1999 21:58:03 +0000 (21:58 +0000)]
Revert previous change -- it was an accident.

25 years agoavoid some NO_<cipher> problems
Bodo Möller [Mon, 2 Aug 1999 21:44:49 +0000 (21:44 +0000)]
avoid some NO_<cipher> problems

25 years agogenerate error message
Bodo Möller [Mon, 2 Aug 1999 21:41:46 +0000 (21:41 +0000)]
generate error message

25 years agofix previous modification -- if ssl->cert is NULL, don't follow the pointer.
Bodo Möller [Mon, 2 Aug 1999 20:09:23 +0000 (20:09 +0000)]
fix previous modification -- if ssl->cert is NULL, don't follow the pointer.

25 years agoavoid cast
Bodo Möller [Mon, 2 Aug 1999 19:55:51 +0000 (19:55 +0000)]
avoid cast

25 years agoThe SSL_CTX's cert structure is not relevant for the SSL
Bodo Möller [Mon, 2 Aug 1999 18:40:36 +0000 (18:40 +0000)]
The SSL_CTX's cert structure is not relevant for the SSL
(because now SSL_new makes a copy).

25 years agoSSL_shutdown was done too early.
Bodo Möller [Sun, 1 Aug 1999 11:19:59 +0000 (11:19 +0000)]
SSL_shutdown was done too early.

25 years agodon't prematurely shut down socket -- use SSL_shutdown
Bodo Möller [Sun, 1 Aug 1999 10:04:37 +0000 (10:04 +0000)]
don't prematurely shut down socket -- use SSL_shutdown

25 years agoExtra i386+gcc bn_div.c tune-up featuring inline division and saving
Andy Polyakov [Sat, 31 Jul 1999 23:27:41 +0000 (23:27 +0000)]
Extra i386+gcc bn_div.c tune-up featuring inline division and saving
the remainder left in %edx. Here is the resulting performance improvement
matrix (improvement as a result of this *and* previous tune-up committed
two days ago). The results were obtained by profiling the "div" part of
the crypto/bn/bnspeed.c.

CPU BN_div bn_div_words overall comment
------------------------------------------------------------------------
PII +16% accumulated by +2-3% PII multiplies damn fast! Taking
inlining multiplication out of the loop
didn't make too much difference.
Eliminating of the multiplication
involved in remainder calculation
is the major factor.

Pentium +45% accumulated by +7-9% mull isn't that fast and replacing
inlining multiplications with additions in
the loop has more visible effect:-)

MIPS +75% +12% +20-25% In addition to the taking mults
R10000 out of the loop (giving 12% in the
asm/mips3.s) three mults were
eliminated in BN_div.

Alpha +30% +50% +10-15% Same as above. But remember that
EV4 bn_div_words is a C implementation.
It takes 4 Alpha mults in C to do
the same thing as 1 MIPS mult in
assembler does. So the effect (50%)
is more impressive. But not the
overall one... Well, if Alpha
bn_mul_add would be implemented
in assembler overall improvement
would be closer to MIPS...

25 years agoby request: let BN_dup(NULL) just return NULL
Bodo Möller [Fri, 30 Jul 1999 19:22:57 +0000 (19:22 +0000)]
by request: let BN_dup(NULL) just return NULL

25 years agoBignum division tune-up. Idea is to move multiplications in front of
Andy Polyakov [Fri, 30 Jul 1999 11:43:43 +0000 (11:43 +0000)]
Bignum division tune-up. Idea is to move multiplications in front of
loop body and replace 'em with addition/subtraction.

25 years agocorrect error signalling for opendir() failure
Bodo Möller [Fri, 30 Jul 1999 10:43:34 +0000 (10:43 +0000)]
correct error signalling for opendir() failure

25 years agoFix to PKCS#12 code to use the cipher block length when allocating a buffer
Dr. Stephen Henson [Fri, 30 Jul 1999 10:11:21 +0000 (10:11 +0000)]
Fix to PKCS#12 code to use the cipher block length when allocating a buffer
for encrypted data, rather than hard coding '8'.

25 years agoAllow the PKCS#7 (S/MIME encrypt) application to support more than one
Dr. Stephen Henson [Fri, 30 Jul 1999 01:12:46 +0000 (01:12 +0000)]
Allow the PKCS#7 (S/MIME encrypt) application to support more than one
recipient.

25 years agoTidy up pkcs12 application.
Dr. Stephen Henson [Thu, 29 Jul 1999 21:50:34 +0000 (21:50 +0000)]
Tidy up pkcs12 application.

25 years agoAvoid path separator problems.
Ulf Möller [Thu, 29 Jul 1999 17:27:29 +0000 (17:27 +0000)]
Avoid path separator problems.

25 years agoNo use in naming the cblock _; the structure still is incompatible
Ulf Möller [Thu, 29 Jul 1999 16:15:48 +0000 (16:15 +0000)]
No use in naming the cblock _; the structure still is incompatible
to Kerberos.

25 years ago*** empty log message ***
Ulf Möller [Thu, 29 Jul 1999 14:43:51 +0000 (14:43 +0000)]
*** empty log message ***

25 years agoobj_dat.h is autogenerated (it was in the CVS because old versions of
Ulf Möller [Thu, 29 Jul 1999 14:10:20 +0000 (14:10 +0000)]
obj_dat.h is autogenerated (it was in the CVS because old versions of
Configure didn't generate the file in Windows builds).