Wyk3d

The easiest I know, but still not very easy, way is to use a profiler like Rational Quantify / callgrind / DevPartner profiler (don't remember the exact name) / gprof etc ..

That is why it's important to have datetime + hash .. you use only datetime in 99.99% of the cases, but when there's a tie you check the hashes.

Did you do a "git fetch origin" before doing checkout ? Remember that checkout doesn't download anything, it just changes to a specific version that you have already downloaded. Fetch downloads stuff. Pull is just a shortcut for fetch + merge.

Some people just chose to turn their face instead of trying to work things out, but there's no point in beating a dead horse .. closed.

I have read a lot of replies and I've seen very few intelligent comments on things that are really bad in git. The vast majority of the issues do have good solutions. It's just that it takes time until everyone gets to know these solutions, so you get a lot of noobs thinking they don't exist. Once you get used to it, it has some features which are a lot better than SVN, even on Windows. I'm using Windows for example, as well as the majority of the dev team, and I was as shocked about this change as anyone else, but I found a lot of cool stuff in it that make up for most of the annoyances. The dates aren't as convenient as revision numbers I agree but they work with identifying and comparing commits and you saying they are pathetic is just you being too stubborn to try using them. Do they ? From what I've heard old SVN patches apply just fine if you make sure your paths are right. It takes about 15 clicks in total to update to the latest commit using the GIT GUI. Even less clicks if you want to update to a specific commit by SHA, though more clicks if you want to search a SHA by date first. Point is .. you can do with just clicking and copy paste if you don't know how to do anything else. There were some issues that were fixed recently, and if you see new ones feel free to report them and they will get fixed. We currently rely on your feedback to fix the problems. If you don't test it then we can test all want with one player in the server, there is no way for us to know that something is wrong with it. We are working on changing that, but it will take time. As I've written a million times, and people who have experience in large projects agree with me, if you do not review the changes and just commit them without knowing the code, you will hurt yourself more in the end. Everyone knows the patch is stable, but you need much more than that to have something integrated well. I don't believe anyone officially said that, or if he has it's not something that is agreed upon by the whole team (there are lots of things like that, we are all different people and can disagree sometimes). There are several anticheat measures in place already in the core. I believe the real issue of why the changes haven't gotten in is because some of these changes are very hacky or too slow or something, don't know actually. It was tested for a long time by a lot of people on all platforms, nobody reported any problems with it, shame on you for not testing and reporting the problems sooner so they could have been fixed before it being committed. It was considered stable but that's not why it was added but because we have a developer in the team who knows how to work on it and fix bugs if necessary. That can not be said about the arena patch, very few of us know how battleground stuff works. If something gets broken later we cannot fix it without reviewing it, so it's better to add it only after it's reviewed.There have been very few real reports of problems with ACE (I mean other than complaints and whining). The one with disconnecting was solved by Derex and committed later on.

You can do exactly the same thing with datetime+hash as you did with revisions. It's a bit harder to remember and use but there is no ambiguity. For example, someone tells you that a certain bug was fixed in a commit at date X, without giving you a hash, so you'd want to update to that. Here's one way to do with just the GUI:- open the git gui and in the last menu select fetch -> origin .. this is only needed to make sure that you have downloaded all the latest stuff before proceeding, will not change any of your files - in the first menu click "view history of ..", this will open gitk - on the right you see the datetime of all the commits, scroll down and quickly find one that matches .. although rev number might be easier, this is still like a dictionary search, humans are fairly good at it .. - the SHA of the commit will appear on the left, select that and copy it - close gitk and in the gui -> branch menu -> checkout -> revision -> paste the sha -> ok (note: all of the above can be done in 3 lines in console with git fetch origin master, git log date..date to get the SHA then git checkout $SHA) Or someone tells you that you need a revision newer than a certain date and you just want to find out if you have that .. just look at the motd in the client and you will see the date and hash of what you are using. Do the same when you want to report bugs to let people know what commit you are using.

For people who just want to download mangos and run it, there's not that big a difference. Sure TortoiseSVN looks better but you do have a Git GUI with explorer integration in msysgit too that works just fine and with a couple of clicks you can clone the repository fetch new changes and checkout one of the branches. Applying patches is a bit harder but you don't have to rewrite anything, the easiest way is to copy the patch just outside the repo and then you can do git apply ../patch-name and you're done.. For more advanced usage the learning curve is much steeper, but once you've understood it you'll never forget it.

I think you can do all those things in the DB, you don't need to write a script for it ..

We have a virtual void OnPlayerEnter(Player *) {} in InstanceData but that's called when a player enters an instance ..

http://github.com/mangos/mangos/commits/master .. all of those commits are newer than what was svn revision 6767.

Until a permanent location for the linux crash dumps are decided please post them in this thread as you did on the old forum.

Until a permanent location for the windows crash dumps is decide please posts them in this thread just like in the old forum.

Did you use freghar's first command (the one with -1) this time ?

One of my favourite git tools is "gitk" which is a GUI that shows not only the revision log with all the changes but also you're local commits and uncommitted changes and it does this a lot faster than the equivalent SVN tools. Things like qgit and tig do the same but I haven't tried them yet.

In fact one of the main reasons for this was to make things easier for the patch contributors on all OSs ..

You could write the scripts in any language as long as you can create a DLL out of it that can be loaded. There was a project that used LUA in the past so it can be done but it was discontinued.

There was a high amount of complaints about SVN, that it's slow, that it's hard to make branches that it's hard to merge changes etc so now we switched to Git. It makes a lot of steps forward and a few back, but once people get used to it's not so hard to use. Just not true, as has been explained above. There are people who say they've been banned for no reason other than speaking up, but they are really lieing to themselves to preserve their image. Some people like Paradox say that when in fact they got banned for insulting other forum members and staff, and just generally being a total pain in the ass, not for the content of what he said. Other people like Brian and KingPin say these kinds of stuff because they made rants, complaints, insults and acted childishly. Their posts got closed because of that and now they're saying we didn't listen to them. About responding to requests there usually isn't even enough time to read all of them let alone respond. In the dev team there hasn't been much of a split. With the trinity thing those who left saw that it's impossible to work with Brian so they either took a break from development or came back. As for Git, all of the dev team has switched over and people are making commits again, there's no split there. A lot of us were disappointed by the poor performance of SVN but other than that this was more for the patch developers who found it hard to merge their changes in SVN. It will take them some time to learn Git but it's supposed to make their work easier. It has been talked about for a loong time that we need to think about git and about clustered forums, but nobody had the time to do it until recently, just a coincidence. It's a learning project but not for people who don't know anything, they should first take courses in C++ and development before they start making patches. After that you can read the code and good C++ style documents itself and there are some comments here and there too. Above all that you can always ask questions.

Technicly you can do the most common operations like checking out the repository (cloning) and updating (pulling) new revisions 100% from the msysgit gui on windows. Other operations require more console work though ..

Wasn't the original plan to have svn + git ? SVN does have some advantages compared to Git besides many disadvantages.. You should at least keep the svn repository around for later use.

Try [6650]+.

Thanks for the info . That's actually good news for us, since it's easier to implement that way. That was not how it worked until now but it's fixed in [6481].

Do single target auras like sheep or hibernate etc get removed when the caster or the target goes through a portal on official ?

Added in [6435]. Thank you! Note: I didn't add the changes in loothandler because I'm not sure how you'd like corpse looting to work in future. It can be added later if you provide more information on that.

<div class='quotetop'>QUOTE (w12x @ Jul 22 2008, 11:50 AM) <{POST_SNAPBACK}></div> Didn't see that one, sorry, my bad.

<div class='quotetop'>QUOTE (w12x @ May 3 2008, 10:42 PM) <{POST_SNAPBACK}></div> And maybe the people who don't want the extra overhead of doing moveinlineofsight for the vast majority of gameobjects that don't need it will get mad at you for NOT spawning trigger npcs. You seem to be doing big changes without discussing it with anyone... IF it's decided that we don't want AI for gameobjects then you did all that for nothing.