4 ms·
> I really don't, at least not on the scale it was a couple of years ago. I expect there are a handful of organisations who are still stuck in IE6 land, but it'
by dpark 15y ago
> I really don't, at least not on the scale it was a couple of years ago. I expect there are a handful of organisations who are still stuck in IE6 land, but it's no longer the limiting factor it used to be.
And we can all be thankful for that.
> I guess I lack sympathy because, ironically, the vendors are still using old and underpowered tools and casual processes for their development, which is why we need so many security patches. Usually, those tools start with a C and end with the word "compiler". Often, the development processes are trendy, "Agile" things that emphasize prototyping and pushing rapid updates over clear specs and systematic designs, and that treat techniques like unit testing as some sort of divine correctness guarantee instead of a back-up to good basic development practices.
I suppose you could use HotJava or Lobo if you want a browser not written in C(++). Really, if you think that Firefox is poorly written and managed, why not move to something else? I feel like software companies are always under fire for using outdated techniques and poor quality control and whatever else, and yet these complaints seem so often to come from people who haven't produced anything on the scale of the software they criticize.
> Well, you reap what you sow. If you don't invest in tools and processes that can build robust software that requires little maintenance, I'm going to bitch if you don't invest in supporting your existing software either.
What projects have you worked on that had 5MM lines of code? Considering how complex a modern browser is, I'm not sure the bug rate is especially high. How many people use Firefox every day and rarely encounter bugs? Seems pretty robust to me.
> We've had programming languages much better suited to application development than C and its derivatives for many years now, and there is really no excuse for still writing things like networking software using a model that has a terrible risk/benefit ratio for that kind of project.
I'm sure you'll be looking into Lobo, then.
> It's also a common fallacy that writing code significantly less buggy than what we put up with today needs heavyweight processes that cost more. There are many people in many industries who do it every day and have the metrics to prove it.
I'm not really sure why you're going down this path, since I don't think I ever said anything about this, but I'll bite. What industries are producing significantly less-buggy code at the same (or similar) cost?
> Sure, but unless people are going to use that bleeding edge technology, why push out new versions of the browser with the corresponding functionality changes? It would be far safer to push only such security patches and bug fixes in existing functionality as are necessary, and make functional changes at a pace that allows for the whole industry to develop sustainably and actually take advantage of the improvements.
Safer? Sure. More expensive? That, too. Maintaining multiple branches is not free. Mozilla has been playing that game for a long time, and I think they're sick of it. I doubt their decision was made indiscriminately.
> Well, so far it seems to have broken quite a few plug-ins and it looks like some of the typography engine bugs are back again, but I've only been using it for two days so I haven't yet tried it with most of the projects I'm involved with.
> In any case, shipping one version without breaking anything is nothing to be proud of. You have to ship every version without breaking anything, and neither Firefox nor Chrome has a good track record in that department.
If they can't ship without introducing functional bugs, then I agree they have issues they need to work out. The plugin thing isn't really acceptable, either. If you're going to have other people build on your product, you absolutely need to support them.
- Silhouette 15y ago> I feel like software companies are always under fire for using outdated techniques and poor quality control and whatever else, and yet these complaints seem so often to come from people who haven't produced anything on the scale of the software they criticize. [...] What projects have you worked on that had 5MM lines of code? I think you know that I'm not going to answer that when I'm posting under a pseudonym, but I've worked on some reasonably large-scale projects. I doubt anything I've written ultimately has as many end users as Firefox, but I'm definitely in both the millions-of-lines club and the millions-of-users club, probably multiple times by now, and I've worked on projects where faults can cause Very Bad Things to happen. > I'm not really sure why you're going down this path, since I don't think I ever said anything about this, but I'll bite. What industries are producing significantly less-buggy code at the same (or similar) cost? My point is that if you're going to write software with a huge cross-section for malware to attack, such as a web browser, and you choose to write it using decades-old technology, where it is difficult to write it securely, then it's hard to take seriously any complaints about how hard it is to maintain that code and keep up with security patches. As for other industries, obvious ones would be defence, aerospace, finance, and infrastructure management. When was the last time you saw a plane fall out of the sky because of software failure, or your electricity supply cut off because the management software for the national grid crashed? (There are also oddities like TeX, but then that was written by Donald Knuth, so it's hardly representative.) Software developers in these industries typically work to different standards than those developing consumer products, and yes, the up-front costs are typically significantly higher. The thing is, by the time you account for the time it takes to fix bugs post-release compared to fixing it at an earlier stage in development, the difference in the total effort isn't such a high factor. And at that stage, you still haven't counted the benefit to society (and to your PR) of not having perhaps millions of users disrupted by each serious bug. The bottom line is that as an industry, we are mostly stuck in the past, but we are there only because of inertia and non-technical factors, not because of engineering merit. As I noted before, this is a rather ironic argument in a discussion about trying to move people onto newer and better browser technologies, but quite apt: it shows how little progress we make if all the better technologies live in their own little worlds. > If you're going to have other people build on your product, you absolutely need to support them. Exactly. I think that's the key point that I and a few other people in this thread have been trying to make.