aboutsummaryrefslogtreecommitdiffstats
path: root/CVSROOT/log_accum.pl
Commit message (Collapse)AuthorAgeFilesLines
* Catch up to projects/peter2003-02-281-0/+2
|
* * Improve a debugging message.joe2002-08-191-3/+3
| | | | | | | * There was too much space (three lines) between the diffs in commit mails. This reduces it to one. Submitted by: fanf
* Populate the branch header in the email with the vendor branch forjoe2001-12-281-2/+8
| | | | imports.
* Remove some trailing whitespace.joe2001-12-131-3/+3
| | | | Submitted by: Jun Kuriyama <kuriyama@imgsrc.co.jp>
* Define $DIFF_BLOCK_TOTAL_LINES to restrict the maximum number ofjoe2001-12-041-3/+22
| | | | lines of diff in a commit email.
* When suppressing a diff because it's too large mention what size itjoe2001-12-041-3/+6
| | | | actually was.
* Add the ability to have diffs inserted into the commit email.joe2001-12-011-0/+73
| | | | Submitted by: jesper
* Move the code that calculates the previous revision number out intojoe2001-12-011-11/+22
| | | | its own subroutine ready for the diff code to use also.
* Add blank lines between subs for consistency.joe2001-12-011-0/+8
|
* Move the map that determines the names of the log files to use intojoe2001-11-301-49/+22
| | | | the configuration file.
* Add a 'cvs-projects' log file for the projects/ subtree.joe2001-11-301-0/+1
|
* Use the template headers from the config file instead of hard codingjoe2001-11-191-11/+3
| | | | them.
* Migrate the code that determines what the name of the committer isjoe2001-11-151-3/+1
| | | | to the configuration file creating a new config variable $COMMITTER.
* Remove all references to /home/ncvs; the CVSROOT env variable is sufficientjoe2001-11-151-1/+1
| | | | | | | | to determine the location now-a-days. Move the name of the exclude file into the configuration file. Partly Submitted by: alfred
* Move the names of some system tools into the config file.joe2001-11-151-3/+5
| | | | Submitted by: alfred
* Show who mail is being sent to.joe2001-11-151-2/+2
| | | | Submitted by: Jon Parise <jon@csh.rit.edu>
* Expand tabs into spaces in the email and log file so that everythingjoe2001-11-151-3/+7
| | | | | | | stays correctly lined up after indentation. Requested by: Dan Langille <dan@langille.org> Submitted by: tobez
* Whoops, reverse the sense of a test.joe2001-11-141-1/+1
| | | | Submitted by: tobez
* Clean up some syntax for clarity.joe2001-11-141-20/+13
| | | | Partly Submitted by: tobez
* Sort filenames so that entries in the generated email are notjoe2001-10-291-2/+2
| | | | dependant upon the order that the committer specified them.
* Make the 'To:' header configurable. (Defaulting to on, and offjoe2001-10-271-2/+3
| | | | for FreeBSD).
* Add a 'To:' header at email generation time.joe2001-10-271-2/+4
| | | | | | Sendmail automatically adds one but other MTA's, like postfix, don't. This should make for a more consistant email across different platforms.
* If a committer modifies more than one area of the tree in one commitjoe2001-10-261-9/+12
| | | | | | | cvs gives them the chance to enter different log messages for each area. If they do then it's really separate commits and so treat it as such, mailing separate emails and writing separate logs for each one.
* We don't fetch the list of email addresses from the temporary filejoe2001-10-261-1/+1
| | | | | that used to be used. Move the bit of code that gets them back into the bit-rotted commented out code block.
* Rip out the final references to internat.freebsd.org.joe2001-10-241-4/+2
| | | | Requested by: will
* Rewrite the email generation routine to feed the email through anjoe2001-10-241-10/+29
| | | | | | anonymous subroutine in the cfg.pm file if it is defined. This allows the administrator to add cvsweb URL's to the email, for instance, at time of generation.
* Clear up the handling of trailing blank lines.joe2001-10-241-3/+2
| | | | | | | All log file entries are now separated by three blank lines; previously only commits were, and directory creates weren't. Email messages now aren't terminated by two trailing blank lines.
* Indent the mail banner by a couple of characters so that it playsjoe2001-10-241-1/+1
| | | | | better with the commitlogs/, and correct the way that its new line is done.
* Move the mail header that carries the list of branches that werejoe2001-10-231-4/+3
| | | | committed to into the config file.
* Back out the last commit. Looking through the cvs source code it'sjoe2001-10-231-3/+3
| | | | | | | not safe to rely on CVS_USER just yet. It doesn't get cleared in normal mode, so it would be possible for a local user to pretend that someone else made a commit in the email (even though the repository itself was correct.)
* Canonicalise the way that we obtain the username of the committer,joe2001-10-231-3/+3
| | | | | | | | | | both in the access controls and in the log message generation. Use the CVS_USER environment variable if it's defined to obtain the true username. This is set by later versions of cvs (it's in 1.11.1p1), and gets around the problem of using pserver username aliases in CVSROOT/passwd. With this fix the committer's true username is checked and reported instead of the system one.
* Avoid getting bitten by filenames that have a ',' character in them.joe2001-09-241-5/+5
| | | | Bug spotted by: alex
* Remove mention of the word 'Attic' from commit mail deltas. They'rejoe2001-09-241-1/+1
| | | | internal to CVS and shouldn't be mentioned here.
* When working out what the previous revision was, for producing thejoe2001-09-241-3/+11
| | | | | delta for a removed file for instance, take into account branch points.
* Add revision deltas for added (new) and removed (dead) files tojoe2001-09-171-22/+99
| | | | | | the commit mail. Requested by: rwatson + others.
* Centralise the generation of temporary filenames instead of havingjoe2001-08-191-30/+27
| | | | it done within each individual script.
* The CVSROOT/scripts use a file, /tmp/#cvs.files.lastdir, to storejoe2001-08-191-1/+1
| | | | | the name of the last directory in a multidirectory commit. Move the definition of this to cfg.pm so that it's defined in one place.
* Add a comment noting that the last directory is written down byjoe2001-06-191-0/+2
| | | | commit_prep.pl
* Add a configuration option to enable the sending of commit mail onjoe2001-05-231-1/+1
| | | | directory creation.
* Migrate some more configuration entries to the config file.joe2001-05-231-48/+5
|
* Finally make the script run under the 'strict' pragma.joe2001-05-221-33/+41
|
* Still working towards this script being entirely strict.joe2001-05-221-10/+14
|
* Remove support for disabling the commit summary in the email andjoe2001-05-221-38/+14
| | | | | | | commitlogs. This was controlled by tweaking the $RCSIDINFO variable, where 0 = off, 1 = mail only, and 2 = both mail and commitlog. It looks like '1' was broken and so not used by anyone, and '0' isn't so useful (I'll re-add support if anyone moans enough ;)
* Use $DEBUG and $TMPDIR from cfg.pm.joe2001-05-201-15/+15
|
* Recently Maxim Sobolev <sobomax@FreeBSD.org> introduced an MFCjoe2001-05-131-0/+1
| | | | | | | | | | | | notification service driven from 'MFC after:' headers in the commit message. This commit adds 'MFC after:' into the rcstemplate and associated code to make sure that if it's used it's used correctly. Based upon code submitted by Maxim Sobolev <sobomax@FreeBSD.org>. Submitted by: sobomax
* Log 'cvs import' reports in the commit log files in CVSROOT/commitlogs.joe2001-05-081-1/+1
| | | | | They've never been logged here before by the looks of things, which seems a little strange.
* Add a couple of missing 'my' declarations.joe2001-04-241-5/+5
| | | | Tidy up a comment.
* Disable 'use strict'. We're not ready for that just yet!!! (It crept injoe2001-04-201-1/+1
| | | | | | from my test environment). Several dunce caps to: joe
* Declare some variables before use.joe2001-04-201-8/+12
|
* Clean up the CVS log message parser and in the process fix ajoe2001-04-201-27/+37
| | | | | potential bug where the contents of the log message could conflict with the parsing process.
* Rename @log_lines to @msg_lines to avoid confusion with an earlierjoe2001-04-191-2/+2
| | | | use of @log_lines in a different context.
* Remove some unnecessary spaces, and add a commented out 'use strict'joe2001-04-191-2/+3
| | | | for later use.
* Tidy up the use of variables in the log message generation code.joe2001-04-191-28/+25
|
* Back out part of revision 1.62 that didn't work.joe2001-02-271-3/+3
|
* Remove the unused $tag variable from the arguments to alljoe2001-02-271-4/+3
| | | | change_summary_X subroutines.
* Apply the style guide to the rest of the subroutines. In particular addjoe2001-02-271-91/+81
| | | | | loads of 'my' declarations. It won't be long before this will run with 'strict' switched on.
* Only add a CVS-branch header to the email if $X_BRANCH_HDR is defined.joe2001-02-271-4/+6
| | | | This allows it to be switched off if necessary.
* Tidy up the formatting using tabs instead of four spaces as perjoe2001-02-231-305/+305
| | | | the style guide.
* Argh! Repeat after me:peter2001-02-041-2/+3
| | | | | | | | NEVER NEVER list cvs-committers@freebsd.org in these scripts by default! At least, not without checking the hostname of the machine first! Pointy-hat to: joe
* More stylist changes. Getting closer to being able to 'use strict'.joe2001-01-291-62/+76
|
* Make the way read_logfile is called more sensible, and perl.style-ify it.joe2001-01-291-19/+24
|
* Apply style.perl(7) guide-lines to a number of subroutines.joe2001-01-291-27/+31
| | | | Remove the 'append_to_logfile' subroutine as it's not used.
* Changes to the cleanup_tmpfiles routine:joe2001-01-291-8/+9
| | | | | | | * local -> my. * Don't chdir to /tmp. * Use $TMPDIR to file the temporary files. * Don't delete the temporary files if we're in debug mode.
* Don't build the final log message onto a list that might not be empty.joe2001-01-161-1/+1
| | | | Problem spotted by: bfumerola
* Fix a debugging format msg.joe2001-01-161-1/+1
|
* Rename a variable: $id => $PID. (Make it clear that this is actuallyjoe2001-01-161-30/+30
| | | | the process id.)
* Smatterings of missing 'my' declarations.joe2001-01-161-27/+27
|
* Make the location of the temporary directory a configuration option,joe2001-01-161-10/+18
| | | | along with the prefix that we use for the temporary filenames.
* Rename some variables so that configuration options are all injoe2001-01-161-9/+9
| | | | | uppercase: $debug => $DEBUG and $rcsidinfo => $RCSIDINFO
* Remove the $freebsd and $crypto variables; they're no longer requiredjoe2001-01-161-6/+1
| | | | now that the header is generic.
* Simplify by making the optional banner at the top of the commitjoe2001-01-161-25/+17
| | | | | mail a configuration variable at the top of the script. (The crypto repository used this, but other projects may find it handy also.)
* Begin the tidy-up to make these scripts easier for third parties to use.joe2001-01-161-22/+30
| | | | | Move the mail command and email recipients to the top of the script where they are easier to modify.
* Dynamically format the changes summary block at the bottom of logjoe2000-08-301-19/+52
| | | | | | messages and commit mails so that the values always fit the columns. Reviewed by: markm
* Add a header to all commit mail showing which branches the changesjoe2000-06-171-0/+10
| | | | | | | | | | were applied to; useful for filtering by procmail, etc. The additional header is of the form: X-FreeBSD-CVS-Branch: HEAD,RELENG_3,RELENG_4 Requested by: andreas & asmodai
* Update the commit scripts to understand 'Approved by:' and kill cvsedit.peter2000-02-131-1/+2
|
* Try and stop the line counter from getting confused by preventing itpeter1999-11-291-0/+2
| | | | | | from seeing any more than the 'lines:' in the rcs header itself. It used to scan the entire log message and could "find" another entry if another log message, complete with full headers, was pasted into the middle.
* Sync with internat.freebsd.org crypto repo.peter1999-09-191-22/+49
| | | | | | | | (Actually, Mark removed a bit of commented code whice I've left in since we might use something like it for using cvs-$branch commit messages one day) Submitted by: markm
* Try and swallow up any extra "/" characters after cvsroot.peter1999-09-111-2/+2
|
* $Id$ -> $FreeBSD$peter1999-08-281-1/+1
|
* Add extra empty line between sets of committed changes for readabilityn_hibma1999-05-271-2/+2
|
* Take perl from /usr/bin rather than /usr/local/bin.jdp1999-04-061-2/+2
|
* Null commit to verify the previous change.. (changes to the cvsroot filespeter1999-02-181-1/+1
| | | | are not active until after the commit that chages them)
* Use a general purpose (read: hack) tool that I've used over the years onpeter1999-02-181-6/+5
| | | | | various things to inject the commit messages. This should give us back the 'From: User Name <user@freebsd.org>' headers we're used to.
* Add missing paren.msmith1999-02-181-2/+2
|
* Update for post-resend changes use. (the majordomo resend on hub ispeter1999-02-181-1/+2
| | | | | using bulk_mailer to inject the mail, the address list is ignored now, this broke the cvs-committers list)
* Minor tidy up.. Update email address. Change an if (0) { .. } to apeter1999-02-161-11/+11
| | | | | | comment-out, and add comments that mlist_map is actually still used (the commitlogs specifically) so people don't keep offering to remove it. :-)
* Test.wosch1998-07-271-1/+2
|
* Strange, I still got the excessive cross-posting errors.wosch1998-07-271-1/+3
|
* Disable excessive cross-posting to cvs-* mailing lists.wosch1998-07-271-1/+2
| | | | This is a workaround until the postmaster fix his scripts.
* Use hostname() to see if we should append @FreeBSD.org to the mailpeter1997-06-091-5/+14
| | | | addresses
* recognise top-level ~ncvs/doc directorypeter1997-05-231-1/+2
|
* Set umask so newly created commitlogs dont get bad modespeter1997-05-171-1/+9
| | | | | | Reviewed by: Submitted by: Obtained from:
* Fix some bugs that I introduced not long before committing last time..peter1997-05-161-7/+18
| | | | | | | | | | | - suppress duplicate copies of mail to mailing lists when directories are crossed. I removed the use of dbm files for simplicity and used flat text files, but I kinda forgot about the implicit unique keys, so work around it now. - Don't log the action header if it's unchanged from the last directory. It could still possibly be duplicated, but I'm pretty sure the branch tag hashing will stop that from happening. (I hope :-) - add an extra 5 chars to the directory column
* RED ALERT! RED ALERT! Mission control, we have a slight technicalpeter1997-05-161-250/+267
| | | | | | | | | | | | | | | | | | | | | | | | | | | | problem... No seriously, this is a major revamp of the log scripts in preperation for an update to a later version of cvs (ie: 1.9.8) Among the changes: - lots of dead code and useless stuff ripped out - lots of things rewritten and/or redesigned - support new message format that's passed from cvs to the log script so that commits to individual files on branches works. - perl5 now *required* due to above (hashes of arrays @{$var${tag}}) - commit message format slightly changed due to per-file branches - new mailing lists (well, from last year) and commitlogs supported, eg: cvs-contrib, cvs-tools, etc - more space for filenames on each line before wrapping. - more intelligent line wrapping, alignment should be better. - no more insanely long Subject: lines. There are more changes in the pipeline, including being able to do a partial commit message when cvs is aborted in the middle of a commit, but since the diff of the changes is now much larger than the file itself, and we need the new cvs support, commit this now. It's had a fair bit of testing in a private copy of the cvs tree, but it's still not all that unlikely that there won't be things I've missed.
* Edit out null `PR:'sbde1997-04-131-2/+3
| | | | PR:
* Don't send mail when somebody does a 'cvs add' on a directory. It's notpeter1996-12-151-2/+3
| | | | | much use anyway and just makes people reluctant to use 'cvs add' even though it's more appropriate than 'cvs import' in several cases.
* Dont forget to clean temporary files when doing short logfile processingpeter1996-03-121-1/+2
| | | | for imports..
* An (ahem) "hack" to stop the duplicate mail messages going to cvs-all..peter1996-02-171-2/+2
|
* Attempt to fix some outstanding bugs/misfeatures/bogons in the CVSROOTpeter1996-02-071-26/+76
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | code, as well as simplifying some admin procedures. The main change is to stop multi-directory commits only having the subject info of the last directory, and not being sent to all the lists, and not being logged to the appropriate commitlogs. updateCVSROOT is gone (yay!), it's functionality is replaced by a cvs-1.6 feature - the 'checkoutlist' file, which is designed explicitly for this. This will stop the easy-import users from getting told the mailing list is out of date when a committer has been added but had missed out on being put on the cvs-committers mailing list. Update the comments and info in the *info files with the cvs-1.6 headers, this fixes a few typos, clarifys some things, corrects some misleading comments, etc. Some stub files added to indicate how new cvs-1.6+ functionality is used, but left commented out (taginfo, cvswrappers, etc) All hard-coded references to /home/ncvs are changed back to $CVSROOT (as suggested to me by bde) now that cvs-1.6+ makes a special effort to ensure $CVSROOT is always set. This allows the $CVSROOT files to be used elsewhere (eg: for testing) with much less pain. mkavail is history as well, it was often being forgotten. cvs-committers now takes the user list from /etc/group:ncvs via a trivial perl script. This is so that we still have the ability to temporarily disable commit access via 'avail' without cutting off the cvs-committers mailing list. Since /etc/group editing is still required, this is not a new admin overhead. If we ever change so that membership of "ncvs" is not required (eg: /usr/bin/cvs becomes setgid and access control is via avail) then we are going to have to do it some other way. (I have ideas) I have tested this on my machines, but it is possible that I have missed something. As usual, please yell if I've botched it.
* Something funny is going on with the mailing list mapping andpeter1995-11-271-3/+2
| | | | | the perl construct "local($dir) = @_; " - perl warns about it, but it doesn't seem to work if I directly assigned $dir....
* Add in revision summaries to the log_accum.pl file...peter1995-11-271-16/+147
| | | | | | | | | I've been using this code in this approximate form for about 6 months now. I'm keen to hear how useful this is/isn't for people. It gives an unambiguous indication of how big the commit is, and the revision numbers are collected into one place to make it easier to find and follow all the parts of a commit over multiple files.
* Correct typo "commiters" -> "committers" as noticed by Satoshipeter1995-11-271-3/+3
| | | | | Also, try a simple sanity check in "commitcheck" to try and head off accidental commits on thud over NFS...
* Bang! Shoot me! I'll RTFM next time, I promise. I thought chop only tookpeter1995-11-021-2/+1
| | | | off newlines. :-O
* Take $ENV{'USER'} for the login name, as rshd, telnetd and rlogind allpeter1995-11-021-2/+2
| | | | | | | | | | set it. I'm still at a loss to explain why getlogin and `logname` (which make a supposedly secure system call) are returning somebody else's username when cvs (a non privileged process) is run on the end of a rsh. If I do: rsh freefall 'print getlogin' it always seems to work... (If this doesn't work after this commit, I might commandeer Jeffrey Hsu's login.. :-) This commit may say 'hsu' though.)
* Have another crack at getting the login name. Try `logname` - it seemspeter1995-11-021-3/+3
| | | | | that the mail process always gets it right and that's a subshell of the perl script log_accum.pl, so it's worth giving this a go.
* For the moment, prefer the getpwuid($<) value over getlogin..peter1995-11-011-4/+5
| | | | | Remove debugging output again.. (which proved that getlogin returns the wrong value sometimes..)
* The following minor tweak adds a tiny bit of debug info so thatpeter1995-11-011-4/+4
| | | | I can see how 'hsu' is getting attributed with some of my commits.
* If any of the three lines:rgrimes1995-07-151-2/+7
| | | | | | | | Reviewed by: Submitted by: Obtained from: Are blank, after this commit, cvs will remove them from the Log: section.
* Okay, I lied, I went and learned some more perl, and hacked thergrimes1995-07-151-1/+9
| | | | | | | | | | Branch: detection logic into this. It's only 8 new lines, but due to the learning curve it took me 3 hours to write them :-(. Also fixed the $FreeBSD$ keyword to be $Id$ since it is turned off. Next.. auto deletion of blank ``Reviewed, Submitted and Obtained" lines. That should only take an hour more :-)
* Hard code path to perl as /usr/bin/perl so we stop picking it up fromrgrimes1995-04-191-2/+1
| | | | | | /usr/local. Submitted by: asami
* Remove the TZ variable from the enironment before calling localtime()joerg1995-04-091-0/+1
| | | | | | | | | to get the commit timestamp. This makes the time stamps comparable. I'm rather in favour for using gmtime() instead of localtime(), but i guess i cannot convince all you american guys. :) I'll let up the RCS checkout to our CVS master...
* Add mkavail script to make a mailing list file out of the avail filejkh1994-09-281-3/+5
| | | | | | | | | | | automagically each and any time it's changed. Change log_accum.pl to send to the new cvs-committers mailing list that uses this file. This should cut down on the size of our monster cvs mail headers considerably! :-) Many thanks to Gary Clark II for these changes. Submitted by: gclarkii
* Escapemetacharacters in directory name so perl doesn't blow chunks on libg++.wollman1994-09-101-1/+3
| | | | CVS:
* Reviewed by:rgrimes1994-05-261-0/+5
| | | | Submitted by:
* Initial CVSROOT setuprgrimes1994-05-241-0/+407
generated by cgit (git 2.34.1) at 2025-01-11 08:57:50 +0800