I've been reading a fair number of the little mini-reviews people have been posting, and the comments in response to them. In no particular order, some thoughts about various things I've seen:

• Session saving seems to be pretty popular with everyone, so I'm glad we got that for 1.1.
• People still think the Acid 2 test matters. I guess I was hoping it had died down and people had forgotten, but no such luck. The most disturbing was a comment someone made that “Safari is a more standards-compliant browser because it passes the Acid 2 test and Camino doesn't” Um... not so much. WebKit is getting better and better at compatibility, but when it comes to actually working with the most sites, WebKit most definitely does not beat Gecko. Large portions of the Acid 2 test are about obscure edge cases that may never appear on any site. The fact that jinglepants did a serious of very target fixes to make WebKit pass Acid 2 was cool, and apparently a huge publicity win, but it did not magically solve all of WebKit's other compatibility bugs. It's like all the hubbub that comes up periodically in the video-card world: it's nice that they can micro-optimize their cards to look good in the benchmarks that everyone uses, but what actually matters is whether or not the hot new games will actually run.
• Yes, we are not on the leading edge of browser features. There were a fair number of “Yawn, Browser X had that feature a year ago”. comments. You know what Browser X has for every value of X I saw in those comments? Paid, full-time developers. The fact that we are staying at least somewhat competetive despite having less that one full-time developer if you add all of us up and all of us being volunteer is, I think, pretty cool.
• Either most people didn't read all the release notes, or they all work for the government. The notes were organized by release, top down, so “New in 1.1b“, then ”New in 1.1a2”, etc. Many, many mini-reviews mentioned Kerberos support as one of the big new features they picked out of those lists—a feature that I think we only ever had requested twice (both by people with .gov email addresses), but was new in 1.1b. So unless there was massive hidden demand for it, its prominence in other people's versions of the feature list suggests a lot of people just never read past the first section.
• We don't have anti-phishing support. This is the one that bothers me, because we never said it, and it's not true. This appears to be people blowing the MySpace password-stealing fix out of proportion; if I have a page on a site you log in to, and I can steal your password without your knowledge, that's not phishing, and that's what we fixed. We'd like to have real anti-phishing as a feature, but we don't yet, and it's unfortunate that people will likely judge us as not having lived up to claims that we didn't actually make.

Of course, that's all smaller, random stuff. The overwhelming tone I saw is that people are happy with the way 1.1 is shaping up. Once we squash a few important bugs, we'll be ready to ship a 1.1 that a lot of people are really going to like.

Category: Camino

Writebacks (0)

With the fairly wide-spread announcement of 1.1 beta, there have been a whole lot more people trying out development builds than the usual nightly build users. With them is coming a flood of emails, forum and software tracker posts, and feedback emails with variations of “I downloaded 1.1 beta and it doesn't work at all”. The problem usually takes one of two forms: CamiTools, or an older version of CaminoSession. But of course, most users just blame us.

This is partially our fault, because there's no extension API in Camino. We'd like there to be one, but frankly we just don't have the time and manpower right now, as getting the browser itself right is higher priority at the moment. So, naturally the people who really want to make extensions are finding other ways. But it's also a significant amount not our fault, because we don't control other people or their code. So if you are one of those people who really wants to write a hack for Camino, here are some guidelines that, if followed, will make us much, much happier. The less time we spend trying to support users having problems with code we didn't write, the more time we have to code. Who knows, maybe we'll even find the time to write an extension system.

1. Don't. Ask yourself: do you really have to write it as an add-on? Camino is open source, and we like contributions. If a feature you want is missing, there's a reasonable chance that we'd like to see it too, and just haven't had the time. CaminoSession is an example of something that I wish had never been developed as an add-on, because we had an open bug for it, and it could have just been built right in. Yes, there are plenty of things we have said we aren't ever going to do in Camino, so this isn't always an option, but please consider it first.

2. Assume nothing. Far and away the biggest headache that CamiTools brought on was the option to use the Metal style for Camino. It worked (during the times that it did work) by shipping a copy of our main nib file with the Metal flag turned on. The assumption here is that we would never change our nib. Meaning, nothing about the UI in the browser window would ever change. If CamiTools had checksummed the original files and only replaced them if the checksum matched, then it wouldn't have been a big deal, since it would have just not been able to enable Metal until a new version was released. Instead it blindly replaced, and if we had changed the nib since the last CamiTools release then Camino just wouldn't open any new windows. Not so fun.

   This applies to input manager hacks as well. CaminoSession called a lot of methods in Camino code with the assumption that they won't change. Several times we changed methods that CaminoSession happened to use, and because it interposes critical methods, when it fails it brings all of Camino down with it. Our number one support request for Camino 1.1 beta has been people who just see blank pages that say “Loading...” because they have an old version of CaminoSession. People forget they installed it, or think they uninstalled it when they haven't, or it just never occurs to them that it's CaminoSession. Camino 1.0.3 works, Camino 1.1 beta doesn't, therefore 1.1 beta is buggy and broken. If you are calling Camino methods, check for all of them when loading, and if any of them are missing, either silently disable or tell the user “Hey, you need a new version”, and either way it'll just be your add-on that stops working, instead of Camino itself.

3. Call a spade a spade. Maybe I'm just old-fashioned, but I really don't like to see hacks being called plug-ins. To me, plug-in implies a supported method of extending an application's functionality. Hacks are not supported. More importantly though, just make it really clear that it's not supported somewhere that users may actually read it. I had at least one user actually arguing with me in a bug that because a nightly build broke CaminoSession it was a bug in Camino, even after I tried to explain. After that, Ben (the author of CaminoSession) helpfully added a prominent note to his download pages, and that did make a noticeable difference (of course not everyone will read disclaimers, no matter how prominent, but that's just the way of the world).

4. [Edit 3/20]: Honor CAMINO_DISABLE_HACKS. Troubleshoot Camino is a helpful tool that we can point people to for debuging that lets them run with a fresh profile. Unfortuntely, input managers and other such forms of hackery don't live in the profile folder, so they can cause problems even with a fresh profile. To make it easier to isolate problems when they happen, please respect the CAMINO_DISABLE_HACKS environment variable; if it's set (which we can easily do from Troubleshoot Camino), just don't load:
   

       const char* disableHackValue = getenv("CAMINO_DISABLE_HACKS");
       if (disableHackValue && strlen(disableHackValue)) {
           // Troubleshooting mode is on; don't do any swizzling
       }

Number 2 is really the key take-away. If you are willing to ride the choppy waters of keeping something working in constantly changing nightly builds, great—just be careful not to sign us up for the added workload too.

Note: To be clear, I'm not trying to hate on CaminoSession in particular, it's just that it's where we learned these lessons the hard way. I think both we and and Ben were broadsided by the problems, as it was new to all of us, and things got significantly better as we started talking more. And I guess that's point 5: come to #camino and chat with us, or email the mailing list. Communication makes all the difference.

[Edited 3/28; I wasn't aware that “haxie” is an Unsanity trademark.]

Category: Camino

Writebacks (4)

Another long dearth of posts, as I've been really busy lately. In very related news, Camino 1.1 beta is out, so get it while it's hot!

In honor of the beta, this installment of my post-silence post-a-day-for-a-week will be Camino-themed. I'll have more to say about the beta itself later this week, but today I want to announce my beta-day present: CookieThief. It's not much of a present, I grant you, but I made it myself, and it's the thought that counts. After I got Safari Keychain integration working and was talking about how I hoped it would help Safari users try out Camino, Smokey pointed out that it would also be nice if there were a way for Safari users to bring their cookies over too, and thus was born CookieThief. Since it turned out to be almost no additional work to make it go the other way too, it's a full Camino <–> Safari cookie sync tool.

Sure, it lacks a disk image, a ReadMe that no-one will ever read, fancy artwork, and other such amenities, but it does its job, and hopefully it'll be one less barrier to trying out Camino. Eventually I'd like to work an initial cookie import into Camino if we can get the UI right, but even then CookieThief might be useful for those who bounce back and forth between the two browsers.

It's not very widely tested, so I apologize in advance if it sets your dog on fire. If it does, I'd definitely like to know about it so I can fix it.

Unless you have one of those annoying little yippy dogs, that is.

Category: Camino

Writebacks (2)