On Thu, Sep 19, 2013 at 1:28 PM, Jan Ehrhardt <php...@ehrhardt.nl> wrote: > "Anatol Belski" in php.pecl.dev (Thu, 19 Sep 2013 20:39:01 +0200): >>that's also good so. If you look at the git history, what might happen is >>that a release is retagged. So even the tarballs were pushed to mirrors, >>they might have to change. The announce is the real deadline where no >>backspace is possible. If the release turns out to be erroneous after >>that, a new one has to go out. > > Suppose a new Wndows-user wants to install PHP. He goes to php.net, > follows the link 'Windows 5.4.19 binaries and source' and finds only > 5.4.20. What do you want him to do? Download a possibly erronous 5.4.20 > or do you think he will go searching in > http://windows.php.net/downloads/releases/archives/ > for the latest 'real' release?
Jan, as I said earlier it is not 100% ideal but it is a matter for a couple of hours. We do not have a technical solution to solve that right now but as soon as there is something ready we will let you know. That being said, I still think it is something really different from using external builds only for the conveniences of having one huge set of extensions in the release. I think that instead of arguing about this problem we should discuss how to share our limited resources and provide everything at one place, using one procedure and without creating extra confusions with custom binaries (that we can't support when bugs are reported, I mean if we can't reproduce a bug for example). Cheers, -- Pierre @pierrejoye | http://www.libgd.org -- PECL development discussion Mailing List (http://pecl.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
