Okay. One bugaboo that several people, myself included, have grumbled about is that despite LJlogin assuring you that you're logged in, ScrapBook refuses to believe it. Well, I've finally done about as thorough an investigation as I can manage, and the results... are not good, and have me rather annoyed.
So what happened to good ol' ljsession? Well, if you log in via login.bml, it gets set to the value of a domain session cookie that ScrapBook can recognize. Problem is, the creation of a domain session cookie can only happen on LJ's servers, because they require a lookup/generation of a time-based randomized "secret" that's stored in LJ's database. Everything else that goes into a domain session cookie, I can construct, but without that secret, I can't actually create one, and LJ's client interface provides for no such thing.
I could, probably, theoretically, do something involving making requests to login.bml and scraping the results I got, but that would be a much more complex, probably error-prone, and ultimately disgusting procedure, so just so everyone knows, I'm considering "ScrapBook login compatibility" to be way the hell down towards the bottom of the feature pile. If someone wants to find someone at LJ to talk to to add a "generate a domain session cookie based on your existing master session stuff" client protocol call, that'd be awesome, but I really can't be arsed to do the necessary "which comm do I post to, and how can I be sure I talk to an Actual LJ Person and not a Volunteer Who Can Only Pass Things Along To Actual LJ People" investigation myself right now, when I have to worry about getting myself organized to work on the chunks of code I can readily improve myself.
Okay. End of tl;dr rambling and ranting.
Not quite an announcement, but more than just silence.
1.2.0 is what I like to call "feature-complete". This means that I've finished all the work of coding in new features and fixing all the new bugs that those new features cause. As it currently stands in my version control repository, LJlogin will, as far as I can tell, perform all the functions I've required of it. (And it's about goddamned time; the last couple of feature-adds involved too much mystery-solving in the debug phase.)
I'm not releasing it yet, though, because there are some re-factoring things to be done based on things I've learned in this dev cycle -- in other words, holy shit, is there some stuff I've been doing in rather too roundabout a manner, and I need to clean that up, or I'll feel dirty about it -- and if I don't do it now, I know I won't bother until the next time something breaks and I curse myself for being lazy. This should, to my mind and in my hopes, not take more than another couple of days at most.
Almost there, folks.
Hi, gang. I've seen all the latest comments in the last post. Yes, LiveJournal apparently did something new in their "security" procedures to utterly fuck LJlogin. Now the ljsession cookie no longer contains the username at all. They've also apparently added an "ljloggedin" cookie containing a subset of the information from ljsession. I'm totally not happy about this.
I've also only been awake for about 15 minutes or so, and have to get ready to go to work. I have no time estimate on a fix. Sorry to disappoint everyone on that score; the last fix was so quick because all it needed was a quick improvement to how I was doing one thing. Now I have to find out if there's any way at all to usefully get a username given the information that's being provided now. It'll take some research, and possibly some support requests or something, and who the hell knows how long it'll take.
Goddamnit. Fuck you, LiveJournal, right in the ear.
As just about everyone knows by now, LJlogin stopped working within the last day or so. I don't know the specifics behind their doing so, but I've tracked down the problem to a change in the ljsession cookie that LiveJournal returns.
It's ultimately my own fault, as the function that extracts the username from the ljsession cookie (used to print the username in the statusbar and to hand to LiveJournal during a logout attempt) made an assumption about the pre-username content of ljsession that was suddenly no longer valid. Thus, the regexp I was using failed, and thus the username wouldn't be returned, leading to much wailing and gnashing of teeth, etc.
In light of this, I'm releasing LJlogin 1.1.1, implementing this fix. It's available at the usual location. This is the only change from 1.1.0; further feature additions are beyond the scope of what I planned on for tonight, although I am thinking of them for the future. Share and Enjoy!
The subject line pretty much says it all. The important stuff, anyway. I did some reading, found where things did and didn't change between Firefox 1.0 and 1.5, and implemented the needed changes. I'll have to admit that I haven't test it myself, having still not gotten my system in general upgraded, but others I've asked to test a release candidate indicated no problems, so I'm comfortable enough going public with it now.
It's available via the link from the userinfo page, as usual, plus the Update button in Tools->Extensions should Do The Right Thing.
Also, for Mozilla Suite users, I'm sorry, but I tried. I really did. There are just too many stupid little differences -- not least of which being the utterly retarded unstandardized "write your own install.js" non-system for installation -- to make it worth the trouble of trying to make it work. I hate to have to put it this way, but suck it up and install Firefox already. :-)
Comments, etc., as usual, are always welcome. Share and Enjoy!
Sorry, 1.5 users, I'm still not ready for you yet.
Anyone still using pre-1.5, however, may be interested in LJlogin 1.0.1, available from the URL in the userinfo and, theoretically, the "Update" button in Tools -> Extensions. New in this point-release version is, I think, fixes for incompatibilities with Google Toolbar and ForecastFox (both of which I've heard complaints about) without breaking something else further.
I may try to follow any guides to upgrading 1.x extensions to 1.5 that I can find, and put up a test release to see if blind crash-course upgrading without a test setup somehow does the job. Or abuse whatever version of Firefox is on the work machine at my new job. Something to get it done sooner than next year, if I can manage it.
As usual, feedback and assistance is appreciated whenever offered. Apologies if this post is a bit erratic; I need to get to bed soon.