• What Are You Working On? June 2015
    1,566 replies, posted
[QUOTE=Goz3rr;48012905]How difficult is it to break Git exactly, most of the time I have no idea what am doing and I have yet to do something that takes longer than 5 minutes of googling to fix or lost data[/QUOTE] I have yet to really break Hg once, so I'd say it's very easy in comparison :v: The few times that I made a mistake repair was under ten seconds by pressing [I]Ctrl+U[/I] / doing a [I]hg rollback[/I]. If for some reason a command is interrupted mid-process you only need to [I]hg recover[/I] and can then continue immediately. I don't think there are any other failure modes Hg can enter besides those two... Subrepo deadlock was a thing (pretty sure that happens with Git too if you remove subrepo revisions by amending and then try to amend the main one), but the [URL="https://mercurial.selenic.com/wiki/EvolveExtension"]Evolve extension[/URL] makes that completely safe once you install it. I also like the default branching and sync models more than Git's as they are a bit more robust. (Hg flow doesn't really involve rebasing, since there's no true benefit to linear history there due to sane unlabelled head revision behaviour. If you want to keep a branch from syncing you mark it as private until you're done, but you can also select specific branches like in Git if you want.) [editline]20th June 2015[/editline] I edit history a lot though (and do other strange stuff like [URL="http://www.bytebucket.org/Tamschi/tq-tools/commits/all"]repository reboots[/URL]), so maybe that's why I keep running into issues with Git.
[QUOTE=Rocket;48012898]git > svn[/QUOTE] mercurial isn't subversion though
[QUOTE=Goz3rr;48012905]How difficult is it to break Git exactly, most of the time I have no idea what am doing and I have yet to do something that takes longer than 5 minutes of googling to fix or lost data[/QUOTE] If you never use --force, I don't think you can break or lose anything. [sp]And even if you did a --force push, there's ways to undo it depending on the situation.[/sp]
[QUOTE=Naelstrom;48013091]If you never use --force, I don't think you can break or lose anything. [sp]And even if you did a --force push, there's ways to undo it depending on the situation.[/sp][/QUOTE] Git is just a bunch of immutable datastructures, so even if you do a --force, the data is still there :v:
[QUOTE=Killowatt;48012879]Anyone in here do any SSAO before?[/QUOTE] [url=http://facepunch.com/showthread.php?t=1401228]Natrox posted a guide a while back.[/url]
[QUOTE=jetboy;48013083]mercurial isn't subversion though[/QUOTE] it's a joke
[QUOTE=Ziks;48012958]The worst I've done is commit while not on a branch, which isn't too hard to fix but it's a little scary for a beginner to be told you have a detached HEAD.[/QUOTE] The worst thing I've done is an interactive rebase 30 commits back on a pull request branch, unknowingly adding every commit to the branch. In CLI it looks fine, [url=https://github.com/naelstrof/maim/pull/44]but on github..[/url]
[QUOTE=Goz3rr;48012905]How difficult is it to break Git exactly, most of the time I have no idea what am doing and I have yet to do something that takes longer than 5 minutes of googling to fix or lost data[/QUOTE] Aborting a commit in progress will probably destroy your entire local repository beyond repair. I don't think that is yet a thing that is fixed. It rarely happens, but in case of power loss this is a real issue. Even though file systems probably manage to save and recover the data, any commits not successfully completed will have destroyed the entire repository, since the data is not in a atomic transactional format. At least, not as far as I remember. If anyone has any updated sources on this being fixed, I'd love to hear about it. This is my main reason for not using git.
[QUOTE=mastersrp;48013200]Aborting a commit in progress will probably destroy your entire local repository beyond repair. I don't think that is yet a thing that is fixed. It rarely happens, but in case of power loss this is a real issue. Even though file systems probably manage to save and recover the data, any commits not successfully completed will have destroyed the entire repository, since the data is not in a atomic transactional format. At least, not as far as I remember. If anyone has any updated sources on this being fixed, I'd love to hear about it. This is my main reason for not using git.[/QUOTE] Wouldn't it be sufficient to simply return to the previous HEAD state? Of course the difficulty of this in practice may be a bit more.. involved :v:
I spent some time learning about how managed DLLs work, and got Rant working in Unreal by making a quick and (very) dirty interface! [img]https://miyuki.impulsh.moe/Skkaow[/img]
[QUOTE=`impulse;48013584]I spent some time learning about how managed DLLs work, and got Rant working in Unreal by making a quick and (very) dirty interface! [img]https://miyuki.impulsh.moe/Skkaow[/img][/QUOTE] To make your life easier. [url]https://www.nuget.org/packages/UnmanagedExports[/url]
[QUOTE=cartman300;48013621]To make your life easier. [url]https://www.nuget.org/packages/UnmanagedExports[/url][/QUOTE] I've come across solutions like this, but I wanted to refrain from editing Rant's code.
[QUOTE=`impulse;48013584]I spent some time learning about how managed DLLs work, and got Rant working in Unreal by making a quick and (very) dirty interface! [img]https://miyuki.impulsh.moe/Skkaow[/img][/QUOTE] You have created a very dangerous and wonderful thing. Please make something awesome with this. [editline]20th June 2015[/editline] [QUOTE=`impulse;48013648]I've come across solutions like this, but I wanted to refrain from editing Rant's code.[/QUOTE] I will look into this for you.
[QUOTE=cartman300;48013621]To make your life easier. [url]https://www.nuget.org/packages/UnmanagedExports[/url][/QUOTE] So what does this do? Does it embed Mono or something in the DLL?
[QUOTE=Rocket;48013730]So what does this do? Does it embed Mono or something in the DLL?[/QUOTE] What? No. There's the .export keyword in .NET intermediate language. [editline]20th June 2015[/editline] It's just C# doesn't implement it. [editline]20th June 2015[/editline] Well it's actually more complicated than that with the .vtentry, .vtfixup, .data and .export keywords. [editline]20th June 2015[/editline] [code] .method assembly static void modopt ([mscorlib]System.Runtime.CompilerServices.CallConvCdecl) SomeMethod() cil managed { .vtentry 1 : 1 .export [1] as SomeMethod // il crap here } [/code] For those interested
[QUOTE=cartman300;48013767]What? No. There's the .export keyword in .NET intermediate language.[/QUOTE] Hey, I didn't know that :c
[QUOTE=Winner;48010217]i have absolutely no idea why i started with python 2.7 rather than 3.x only started a week or two ago so it's not like i would have anything to relearn... next project i guess [editline]20th June 2015[/editline] also apparently centos relies on 2.6.6 for yum and other things and will break if you update it rather than installing it side by side fucking centos what the hell[/QUOTE] Install Debian. Centos is RHEL, which is not designed to be updated. RHEL is so your company can write some software for one version of RHEL (or centos) and be extremely confident that it will run on any other install of that version of RHEL. This is great for reproducible builds and uniform deployments but shit for general use.
[QUOTE=Lexic;48013814]Install Debian. Centos is RHEL, which is not designed to be updated. RHEL is so your company can write some software for one version of RHEL (or centos) and be extremely confident that it will run on any other install of that version of RHEL. This is great for reproducible builds and uniform deployments but shit for general use.[/QUOTE] Or Fedora, which is meant to be the desktop version of RHEL (well, the other way around).
Wait hang on if you guys are going to have a source control discussion let me dust off the popcorn maker first
nige if you decide to start over, give either me, [url=http://facepunch.com/member.php?username=ryankingstone]ryan[/url], or [url=http://facepunch.com/member.php?username=polkm]polkm[/url] a yell
[QUOTE=Nigey Nige;48013895]Wait hang on if you guys are going to have a source control discussion let me dust off the popcorn maker first[/QUOTE] cvs > git [editline]20th June 2015[/editline] did anyone actually use cvs ever
I'm messing around with UnmanagedExports in Rant. [IMG]http://i.imgur.com/mTOInHr.png[/IMG] I'll have something usable in a little while. Can someone help me test this when I'm done? I'm afraid I don't have much experience with C++. [B]Edit: [/B]​I pushed a basic set of exports to GitHub. Let me know if they work.
blub blub [video=youtube;Q7KL5Evt-1o]https://www.youtube.com/watch?v=Q7KL5Evt-1o[/video] [sp]don't really know what to do with this cause I don't think my original idea will be very fun[/sp]
[QUOTE=Pelf;48014812]blub blub [video=youtube;Q7KL5Evt-1o]https://www.youtube.com/watch?v=Q7KL5Evt-1o[/video] [sp]don't really know what to do with this cause I don't think my original idea will be very fun[/sp][/QUOTE] honestly just make it an exploration game because that was the most relaxing 47 second video i've seen all week
Do any of you guys have any experience with GTK? The docs are a mess
Berkin managed to figure out how to use Unmanaged Exports with Rant, so you can use that now if you want. It'll be out officially in the Rant 2.0 release, but if you want, [url=https://github.com/TheBerkin/Rant/tree/dev]just grab the dev branch[/url]. Just compile Rant (in x86 or x64), add Rant.h, Rant.dll and Rant.lib to your project, and Rant away! [url=http://hastebin.com/yiketoxuwo.cpp]Here's an example[/url] using my rudimentary C++ skills.
That's fantastic, thanks.
Finally Have a demo out for my first real platform game, I started working on it because the gore system I wrote was to neat not to use. [IMG]http://i.gyazo.com/4fb3ddd0c391fa09620c154b2d120106.png[/IMG] [IMG]http://i.gyazo.com/eb986ed1716373369928dfa5d13f3e82.png[/IMG] [url]http://ibald96.itch.io/kill-all-melons-indev[/url]
[QUOTE=Rocket;48013961] did anyone actually use cvs ever[/QUOTE] I've done things I am not proud of :( ( firmware company )
[QUOTE=Profanwolf;48013336]Wouldn't it be sufficient to simply return to the previous HEAD state? Of course the difficulty of this in practice may be a bit more.. involved :v:[/QUOTE] If the repository binary blob becomes corrupted, then git will not know what HEADs there are, what commits have been made, and so on. I personally use Fossil to avoid this, as it simply uses a sqlite3 backend database system, allowing any unclean commits to be rolled back. You'll lose the latest commits (commit small, commit often), but you'll still have the entirety of the repository history. Obviously all of these are susceptible to filesystem corruption, but that is why you use CoW filesystems to make sure that the same techniques are used by the file system (atomic transactions are a common feature of CoW filesystems = saved successfully or not saved at all). [editline]21st June 2015[/editline] As a sideeffect of this, there's no need for any special library to interface with fossil repositories, other than the library used as backend. Currently, the official one is SQLite3, so you can just use SQLite3 bindings in any language or even the SQLite3 cli to interface with repositories outside of fossil. [editline]21st June 2015[/editline] Additional info on the subject: [url]http://www.fossil-scm.org/index.html/doc/tip/www/selfcheck.wiki[/url]
Sorry, you need to Log In to post a reply to this thread.