home *** CD-ROM | disk | FTP | other *** search
- Changes for UNPOST Version 2.0.2
- --------------------------------
-
- 1) Fixed minor bug with zero length lines.
-
- 2) Added if-defs around sys_errlist declarations so that they
- can be conditionally not compiled.
-
- 3) Modified part 0/# processing so that description articles
- are written to the incompletes file ONLY if the -d switch is
- set on the command line.
-
- 4) Added version printing switch, -v.
-
- 5) Added AIX make file.
-
- 6) Added an NNTP client for automated extraction of all articles from
- specified news groups.
-
- The source code is in the nntp directory, and it can be compiled
- by typing:
-
- gcc -o client -ansi client.c
-
- on a Sun running:
-
- SunOS 4.1 4
-
- using gcc:
-
- version 2.1
-
- Documentation for CLIENT is in the nntp directory.
-
- This program is NOT portable. As far as I know, it can only be
- used by the people who have, at a bare minimum, a Unix that
- supports sockets as used in the program for communications.
-
- If the above does not make sense to you, don't even try it,
- unless you have a Unix/C/Socket/BDS/Communications expert to
- try to help you.
-
- CLIENT is provided as a tool only for use by experts. You have
- been warned.
-
- Changes for UNPOST Version 2.0.3
- --------------------------------
-
- 1) Fixed bug in the way part 0/# segments were handled, as they were
- not being written out properly.
-
- 2) Fixed bug in the way part 1/# segments with no uuencode begin line
- are handled. An error check was placed in the decode logic to report
- an error if the file name is missing, even though all the
- segments are present.
-
- 3) Fixed def.cfg so that it would get through the new parser
- (removed unnecesary commas).
-
- 4) Removed '^From ' as one of the segment prefix recognition regular
- expressions from the default configuration. Sorry about that,
- for you people whose news readers (like nn, I think) use that as
- the first line, and who process email.
-
- To make up for it, a configuration file (called email.cfg)
- has been included that should make UNPOST act properly.
-
- 5) Added an OS/2 EMX/gcc makefile (called makefile.os2) that has the
- necesary define set to compile out the sys_errlist declarations.
-
- 6) Added a summary file (you're reading it now), that contains the
- latest changes as well as a summary for use in posting a
- description and keeping providing a location for only the changes
- that have occured since the last version. All changes from
- version to version, from now on will be copied to the changes.doc
- file.
-
- 7) Added a syntax check and very brief syntax description line to
- retest.
-
- 8) Added a new debugging utility, mofs. It is a single file program
- that allows the user to view a file, ala more, starting at a
- decimal offset.
-
- 9) Added a configuration file for use in UNPOSTing comp.binaries.ibm.pc
- and comp.binaries.ms-windows postings (named cbip.cfg).
-
- 10) Added a new command line switch (-b) to tell UNPOST to write the segments
- of an incomplete (or damaged) posting to separate files, instead of
- to the incompletes file.
-
- Changes for UNPOST Version 2.1.0
- --------------------------------
-
- 1) Bug Fix: UNPOST would not recognize a short (does not start with a
- 'M', ' ' or '`' character) UU encoded line as the first uuencoded
- line of a segment. Major bug (truncates files), rarely happens
- (UNPOST will only skip this line if it is the first line in a
- segment, and that only happens when a segment is posted with only
- three lines) therefore hard to notice, but very easy to fix.
-
- 2) Bug Fix: UNPOST would add one or two extra bytes at the end of a
- file when decoding. This is due to uuencoders that put out four
- characters irregardless of whether or not all three bytes are
- valid. I've been told that this is the standard, so UNPOST was
- wrong (BZzzt! Got me!).
-
- 3) Modified the email.cfg file so that it is valid for both the nn
- news reader and email extraction. Thanks to the alert user who
- not only found the problem, but also designed THEIR OWN CONFIGURATION
- FILE (yes, it can be done!) and sent the change to me.
-
- 4) If a segment has multiple sources of information for the part # of
- # of parts and binary ID information, the last source now wins. The
- last source is usually a header dumped into the body by an auto
- uuencode and split program, and this information is more likely to
- be correct than the hand typed Subject: stuff in the header.
-
- 5) Tacked the Subject: line tree into the body, as well as the header
- in the default configuration. This is for the new anonymous service
- postings that put a copy of the header from the original message
- into the body of the posting (irritating, to say the least, that
- the two Subject: lines often disagree).
-
- 6) Moved all compiler header files (.h) files into a file called
- "compiler.h" to simplify making changes as new systems are brought
- up, and for things that are not the same across all systems.
-
- I hated doing this, but it should make things a little easier
- for those who know enough to make some minor changes to port
- it to their systems. This suggestion was from Tom Lane
- (Independent JPEG group, check out the djpeg and cjpeg code,
- this stuff is really useful to have around).
-
- 7) Made a few changes according to the suggestions sent to me by a
- user who modified the program slightly to run under VMS.
-
- 8) Added if-def's for nn vs. rn news readers, and defines for setting the
- binary switch default settings. See file compiler.h.
-
- Another Tom Lane suggestion.
-
- 9) Changed source slightly so that the SEGMENT begin line will also be
- checked for part # of # of parts and binary ID information. This
- change was made to assist me in creating a working configuration file
- for Fido Gateway and UUNET users.
-
- 10) I added an example to the Theory of Operations section in the
- documentation, and moved some other sections around as per request
- by a user who thought the order of the configuration, configuration
- file and regular expression tutorial sections was confusing.
-
- Thanks for the input.
-
- 11) Wrote my own MemCopy and MemMove functions to quash the idiocy with
- having to if def the hell out of the code to decide which one of
- three functions to use. Also, I have exactly the same version of
- gcc on two exactly similiar machines (both the same type of Sparc
- Station, running exactly the same version of the operating system)
- but on one machine, gcc bitches about conflicting definitions for
- memcpy between the builtin function versus the header file, while
- on the other machine. . . silence. I really hate this sort of
- garbage.
-
- Hey, Tom! Give me solution for this one, and I'll implement it
- without expressing any opinion at all. . . :-)
-
- 12) Changed the routine that extracts the binary ID string to use only
- one word (a string delimited by white space) out of the string,
- preferably one that has a '.' character in it. This string is
- filtered to have only MS-DOS/USENET standard file name characters
- in it. Hopefully, this will reduce the number of misinterpretations,
- and avoid putting segments of the same posting into different
- file segment lists. It should, for example, solve the problem
- of putting
-
- Subject: WHITE.GIF (Part 1/2)
-
- and
-
- Subject: WHITE.GIF, part 1 of 2
-
- into two separate file lists, as the comma is NOT an acceptable
- file name character.
-
- Changes for UNPOST Version 2.1.1
- --------------------------------
-
- 1) Bug fix. On MS-DOS and OS/2 systems, if you feed UNPOST a non-text
- file (one that has only line feeds, no carriage returns), very
- strange things happen. The simple fix was to change from opening
- the source files in text mode to opening them in binary mode.
-
- 2) Change 12 for version 2.1.0 had a bug in it that in some (rare)
- circumstances, would cause core dumps due to accessing a NULL
- pointer. This has (fingers crossed) been fixed. Thanks to the
- user who analyzed the problem and sent me the article header
- that was causing the core dump.
-
- 3) The help summary now tells the default state of the binary switches.
-
- 4) In response to the people who need support, a small amount of
- debugging code has been added. This is activated by defining the
- UNPOST_DEBUG variable (in compiler.h).
-
- 5) New switch (-r) tells UNPOST which one of four news readers you
- are using.
- -r r For RN, -r e For Email
- -r n For NN, -r g Look for Newsgroups: line.
-
- 6) For OS/2 only and if the proper variable is defined, UNPOST will
- set the file name munging switch on the basis of the file
- system type of the default drive.
-
- jstevens@csn.org
-