freghar

It has been noticed earlier, I plan to monitor such users and if they won't prove any non-bot like answers, they'll get deleted. edit: 5th message proved it was a bot, so .. deleted and banned.

phpmyadmin is fine if you have an updated version ... or you might use direct DB access with some GUI client such as navicat / sqlyog. The thing is not to rely on web app only, use mod_evasive on apache, use port knocking on firewall to get dynamic access, etc, etc ..

Yes, firewall.

http://milw0rm.com/exploits/8921 - 2009-06-09 http://milw0rm.com/exploits/8992 - 2009-06-22

I agree with that, my "playing smart" was just a sarcasm on "try to optimize our data in the Big Blob ourselves". I found something from 2006: so it's maybe a good idea to split some to a separate table anyway .. in the long term.

Oh .. so I probably missed something. Many sql servers try to buffer/cache something when they are able to (and config allows them to) .. also a storage optimization is in place. All those things doesn't apply on text fields, because the server can't know what does the field actually contain. A page in a book? A list of bitmask values? It cannot guess if the field can be fragmented for better performance or cached in parts. When you split the field into parts which are clear for the sql server, it can handle them better. Imagine someone writing tons of volatile asm() in C just because he thinks it will outperform the same code compiled by a C compiler, which isn't true in many cases as the compiler is aware of many things the programmer might not be. As for the size (I already wrote that somewhere) - a string of three digits (+ whitespace) is equivalent to normal 32-bit integer in size, both have 4 bytes. If the number has less than 3 digits, it's smaller in it's decimal string representation, if it has more (say .. 4 bilion), it's much smaller in it's normal form. The question is whether we play the smart guys and try to organize our data blob or leave this work to the sql server. I'd personally prefer the second variant. As for the first one - we could use a simple text file for storing everything instead of database.

Of course you can. Using a firewall for example. Or ACLs, ..., ...

uh ... 8-char bruteforce on an MD5 hash (local, no network) could take almost half a year on my old 2.6GHz Celeron, so I really doubt he used that, my cracking tool (I believe it was jon the ripper) was able to do few milions of iterations per second (IIRC) while your man would unlikely get 10000+ / second.

,,, and the pretty old http://repo.or.cz/w/mangos-git.git?a=tags edit: for the git part .. you can abbrev the hash and use git-log style of specifying ranges, ie. c288007..a58c1cf for 243

Lua ofc needs to be compiled before it can be executed. This process can, however, stay out of user's attention, on script reload for example.

I'll post it here and update from time to time "5" WARNING: Those commands in the current form won't work always (for experienced users with cutomized repository) and can be used in a more powerful way, but they should work fine for those whom this FAQ is intended. ------------------------------------------------------------- "4" git reset --hard 123456789 where 123456789 is a commit hash. You can get the hash on http://github.com/mangos/mangos/commits/master or http://repo.or.cz/w/getmangos.git?a=heads by clicking at the commit and looking on the URL for example. ------------------------------------------------------------- "4" In most cases no, delete everything EXCEPT the .git directory and then run git reset --hard origin/master (or replace master with the branch you want). ------------------------------------------------------------- "4" Run git gc once a month. ------------------------------------------------------------- TODO HERE

http://repo.or.cz/w/getmangos.git?a=commitdiff;h=e4fea4581 -- should be okay now

Created tag object acdb1cb called v0.13 pointing to commit object b29aa51 ([7986] MaNGOS 0.13 release.) with a note. $ git tag -ln v0.12 MaNGOS 0.12 v0.13 MaNGOS 0.13 -- moving from client 3.0.9 v0.13-dev1 MaNGOS 0.13 dev1 -- moving from client 3.0.3 v0.13-dev2 MaNGOS 0.13 dev2 -- moving from client 3.0.8a

git log -1

It's simplier than C++ and designed for "scripting", widely used amongst tons of projects, from games to window managers. It has more power than a config file and it's less error-prone than C/C++. Some projects use perl/python as scripting language, but LUA wins statistically due to it's simplicity and specific usage (which includes good integration within other languages).

What do you mean? My point was that many people just click the "fork" button and doesn't care anymore. So they create a new useless fork.

Perhaps if there weren't too many useless forks (I guess 90% of them haven't received any new objects which aren't on mangos), it would make the graph update faster ..

1) take a look how .announce command work (Level1.cpp probably) 2) use that function in the .ban command handler (Level3.cpp I guess)

Hi. ... if you say so .. @ my promised article I mailed you about: As I told Vladimir on IRC, it got deleted somehow on my HDD, I had it both written and corrected + did defragmentation that day... Furthermore I don't have the mood of writing it again, not anytime soon, so perhaps some other time, sorry .. edit: I wonder .. what happened with buildbot?

Just a note; the "# git pull git://github.com/mangos/mangos.git 310" is useless in your script ...

or git add . git diff HEAD > file.diff

automake: change your CC variable ... or do something like ./configure --this-bloody-mess --enable-cli --etc CC="my_compiler_binary_name" ie. CC="gcc"

Hello, just a few suggestions: * apt-get is obsolete in Debian, I don't know how that's in *buntu, I would use aptitude or such more advanced tool, but whatever, apt-get is the most error-prone solution here .. * Use one command per line, no semicolons! If an error occur, the command list continues executing and thus you're probably unable to find out where the error occured. If you absolutelly _must_ use several commands per line, because you feel it's somewhat cool, use `&&' instead of semicolon. This will most likely stop other commands from executing if an error occur (due to return value). * I'm really surprised you used git-am. Almost noone realizes that I made SD2 patch usable for git-am. You should however include a git-apply version (or at least mention) for people who don't want any custom commits, because they git-stash all their custom stuff befure update (to make it fast-forward) .. * just btw .. you can use wildcards even like "git-am src/bindings/ScriptDev2/patches/MaNGOS-*.patch" to get only the .patch file (no gzip) ... or use gzipped one together with zcat. * Delete everything in the autoreconf line except "autoreconf --install --force", everything else is already done by this one command. * I don't know why most guides use "--enable-ra" even if 95% of the people using this guide won't ever use RA. * Again, separate make from make install with an error check, your current solution is really, really bad as make can likely fail (which isn't true for mkdir for example). * Using mysql passwords on command line is rather dangerous, better solution would be to use "source" from mysql client or doing something like "cat file1 file2 file3 file4 | mysql mydatabase -p" and typing the pass once. But that is, of course, not so nice, because you won't know which file caused a possible error. edit: * win-SCP would be perhaps easier than filezilla, I don't know if filezilla runs without instalation .. and if remote server doesn't support sftp, but it does have scp, you are able to use that.

Just to add - the log is stored locally (it's generated every time from commit parents, it isn't a text file). You can get the "git directory" by using "git clone" as usual ... or try asking in a better way --

What exactly do you mean by "logs" ? The "git log" command should show the log ...