Jarno Mikkola Posted July 27, 2016 Share Posted July 27, 2016 Downloads... ready to hold as ransom. For a cheep plastic monopoly penny... Erhm, could you make an "Release" from that, as the download is just a Master -copy of the files, so it's not an actual "Release" (candidate). I don't have the technical knowhow from GitHub, as I don't use it... but it should be easy to make. Yeah... you might take as much time with that as you like... it's not a critical-here_and_now_problem, but when you will update the files, it can/will rear it's head. The reasoning was put nicely by abg1: Master branches are not always stable. We want to minimize install problems for users. Taking latest version only gives you control over what BWS will download while simplifying maintenance for us; plus you don't have to be as careful about testing your master branch changes before committing them. Plus you don't need to change version for every commit - only for releases. Link to comment
Jarno Mikkola Posted July 27, 2016 Author Share Posted July 27, 2016 Jarno, there's no difference, download the zip all the same.Then what do the Github files do in the "EET-Master.zip" archive ? The one called ".gitattributes" and the other. AND when K4thos desides to upload a file to the master folder... say a .tpa file and it requires a .tpb file that he has not uploaded yet... and someone downloads the Master archive... they get an incomplete download ... while a Release Candidate, where a candidate made, the process is not automatic for the release candidate download because the archive has to be separately "made", just like the first time. This is the reason why the GitHub has the function to make a "releases", instead of just "masters". Link to comment
K4thos Posted July 27, 2016 Share Posted July 27, 2016 Then what do the Github files do in the "EET-Master.zip" archive ? The one called ".gitattributes" and the other. AND when K4thos desides to upload a file... say a .tpa file and it reguires a .tpb fiel that he has not uploaded yet... and someone downloads the Master archive... you get an incomplete download ... while a Release Candidate, where a candidate made, the process is not automatic for the release candidate download because the archive has to be separately "made", just like the first time. ? I wasn't aware about such problems. From my experience it's a matter of clicking "Clone or download" and than "Download ZIP" and I always got full package. Don't worry about those 2 junk file named .gitattributes and .gitignore - they can be deleted. Link to comment
Jarno Mikkola Posted July 27, 2016 Author Share Posted July 27, 2016 I wasn't aware about such problems. It's not an actual problem, but a matter of a convenience. What I mean is, you have the source of the release, not the actual Release (Candidate). Aka you make us download a folder like this for BWS. There's distinct negative connotations of not making actual releases... while saying "this is the one". Link to comment
lynx Posted July 27, 2016 Share Posted July 27, 2016 But it's the same, just those two extra files that have no effect on users. If you assume he is a shitty committer, why would snapshots be any better? They're also a click a way to generate and are generated from the same files. There's no manual cherry-picking. Link to comment
Jarno Mikkola Posted July 27, 2016 Author Share Posted July 27, 2016 But it's the same, just those two extra files that have no effect on users. If you assume he is a shitty committer, why would snapshots be any better? They're also a click a way to generate and are generated from the same files. There's no manual cherry-picking. Well, ask why the last 7 post in this page were made, by Argent77 from him... I bet he make a little more than a 1 file change per each of those updates/releases... cause he could. Even though they are all snapshots. PS, the posts are made in the same day, cause the forum was down and the backups he had made were removed, so the posts are re-posts of the old changes, if you remember. So don't look at the posts date/time. Link to comment
argent77 Posted July 27, 2016 Share Posted July 27, 2016 NI is different as I have to compile executables for every release. For mods there is nothing wrong to download the master branch directly, especially if you add new commits only very rarely. Link to comment
lynx Posted July 27, 2016 Share Posted July 27, 2016 But in none of those he missed any files, not even earlier (check the log). There was no partial work. Link to comment
Jarno Mikkola Posted July 27, 2016 Author Share Posted July 27, 2016 Yeah, but if the files in the master are being updated, while someone downloads it, you can run to all sorts of issues. It can be a second... that's it all takes to corrupt someones game. Which is why you update the master which is hidden usually... while have a release candidate to point towards for people to download the file compilation they are meant to receive. If you'll notice, the weidu mods have this wonderful thing called version flag. Which you should use and update when you make releases. Even if they are snapshots ... Link to comment
lynx Posted July 27, 2016 Share Posted July 27, 2016 No, git is "atomic" that way, there's no chance of anyone downloading only part of a commit. This is not copying files like on a disk or over ftp, it has its own protocol to avoid such nonsense. It's a multiuser system, this is one of the first thing they had to think about ... Link to comment
K4thos Posted July 27, 2016 Share Posted July 27, 2016 https://github.com/K4thos/EET/releases No problem with releasing it like this Next RC versions will have full changelogs. This time there were simply to many changes related to SoD and patch 2.x to list them all (nah, I was just lazy ). Link to comment
Jarno Mikkola Posted July 27, 2016 Author Share Posted July 27, 2016 ...Thank you very much, now I can stop talking to a man that has never released a thing, even if he calls it GemRB v0.8.4 ... Why I asked this... cause ALIENQuake would have asked it anyways, mostlikely. Cause it's more convenient for him .. Link to comment
lynx Posted July 27, 2016 Share Posted July 27, 2016 This whole git mess should be split out of the topic. Jarno: probably not, as he has an odd stance for a packager: preferring latest code even if untested. Makes BWP even scarier. And the wrapper I wrote for him supports all three options. Link to comment
Jarno Mikkola Posted July 27, 2016 Author Share Posted July 27, 2016 lynx... take this with a smile... would you. T--- And the wrapper I wrote for him supports all three options. So you are the one to blame for ALL the broken and bad installs... Aka it's clearly not your fault ! But it's far more preferred (at least in) concept to only download the releases rather than the masters... unless the release is broken/not available and so the master has to be the one to download. This is clearly what he has always asked ... from what I have read from his posts. Yeah, sorry for derailing the topic ... Link to comment
Mike1072 Posted July 28, 2016 Share Posted July 28, 2016 Split from the EET 1.0 Release Candidate thread. Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.