Community
Participate
Working Groups
While hacking in support for bug 359080 I noticed some differences in the life cycle of the git clone page from the ones I'm used to. - the bootstrap code is not in a dojo.addOnLoad whereas I think it is in other places. I don't know if this matters or not. - the command contributions were being defined before the commands themselves were ever defined, so the contribution code couldn't find the command being added. This didn't cause a problem until a URL binding was added (a keybinding would have had the same issue). In general, the commands need to exist and be added to the service, before being referred to in a contribution, or else the keybindings and such won't be able to find the command info. - updateNavTools is rendering commands by hand vs. using globalCommands.generateDomCommandsInBanner. This means the git code has to know a lot of particulars about how globalCommands is arranging everything. I think instead it should be calling hte global commands code.
I think updateNavTools is worth looking at/cleaning up so that the git pages will be resilient to changes in the page toolbar layout, id's, etc.
Susan, git-clone will be removed in .5. Are those comments valid for the new gir-repository.html?
(In reply to comment #2) > Susan, git-clone will be removed in .5. Are those comments valid for the new > gir-repository.html? The bootstrap issue is still valid. We need guidance from Simon and to make sure the pattern is known/used the same. git-repository.js is bootstrapping outside of a dojo.addOnLoad. compare this to what navigate/table.js is doing. The command issues can be worked later, I'm going to do some API simplification in bug 370014 and once that is done, I'll change clients and ensure things are happening in the right order, etc.
I think that happened during .5.
