|
I work in the lodging industry, specifically in arranging travel. Surprisingly, a lot of hotels still use fax machines as a major source of communication - even the big chains like Choice. We own some hotels too, and I know for a fact that some of them use IE6 because I get the odd unhandled excpetion in my email from the (4 year old!) web apps I built for them. I'm in the middle of rearchitecting our main "get reservations assigned, booked, and billed out" app. It's actually a pretty gigantic WinForms/ADOld.NET application, and we're making a web application to replace it. We can put all of our employees on Chrome, but our customers (who are often low-tech trucking and utility companies) and the hotels we deal with would also need access. The great thing about the web is that you don't need to lock those users out and tell them to gently caress off in most cases. As far as I can tell, progressive enhancement is still relevant. Build a core experience that gets the job done. Then, bring in as many fancy JS modules as you need. It might look janky and be lame in old browsers, but it will work. The kinds of users who are still on IE7 don't give a gently caress so long as they can do what they need without having to download a whole new browser. That's just good web development though - I could be just as well be advocating for blind users. Knowing your audience is #1. glompix fucked around with this message at 19:47 on Jun 18, 2013 |
#
¿
Jun 18, 2013 19:41
|
|
|
|
| # ¿ May 8, 2024 21:54 |
|
|
For me, the biggest issue with testing in IE9+ F12 tools like that is Javascript performance. People are pretty good about dealing with visual quirks at this point, but man those old IE browsers performed like poo poo. Emulated IE7 is a ton faster than real IE7, so that's going to skew you too. If you really have to be lazy for whatever reason and not use a VM, use IETester. It sucks too, but it's not going to mislead you as much in my experience. glompix fucked around with this message at 16:04 on Sep 12, 2013 |
|
#
¿
Sep 12, 2013 16:02
|
|
|
edit: I'm wrong. Thought it was just simply green channel missing, but the text would still be black if that were the case. Looks groovy, though. I'd keep it. 
glompix fucked around with this message at 21:04 on Sep 13, 2013 |
|
#
¿
Sep 13, 2013 21:02
|
|
|
What's the consensus on broadly denying the less-useful HTTP verbs like OPTION and DELETE? Our apps aren't RESTful or anything, just GET/POST. I thought this would generally be a bad idea, but I'm not sure if my idea is valid anymore. I certainly don't like the idea of denying HEAD.
|
|
#
¿
Sep 27, 2013 14:42
|
|
|
Scaramouche posted:I don't know your specific application but I do know that PCI Certification/McAfee Secure/etc. will insist that you get rid of verbs like TRACE, DEBUG, and OPTIONS before they'll clear you for security certification. Thanks! I'm not doing REST/CORS, so sounds like we might as well. I figured some UA or proxy server out there probably uses HEAD, but I didn't know about the compliance stuff since my company is easy street.
|
|
#
¿
Sep 28, 2013 00:57
|
|
|
Cicero posted:Ok, so I have on-site interviews with Facebook in a few weeks. It's for an Android development position (and I'm an Android developer) but I'm told one interview will be about web design. Not like, UX design, but stuff like web development patterns, technology stacks, databases, distributed systems, how to scale, etc. My knowledge of these areas is extremely limited, are there any good resources for this kind of thing where I can learn the basics in a hurry? Check out this site called High Scalability. Read up on eventual consistency for sure. I'm not sure what they'd ask about stacks (maybe just be familiar with something like Node, MVC frameworks. Cache - both on the front-end and backend - would certainly come up. Maybe know something about key-value stores (Redis, Memcached) and how you'd use that to reduce load while maintaining some consistency. CDNs might make sense to review. Hope that helps, not sure what your comfort level is. Good luck!
|
|
#
¿
Dec 12, 2013 19:18
|
|
|
WhiskeyPete posted:Please educate me on how this is possible. There's some technique you can use to make True Color GIFs. The file is probably built on a similar principle. Many frames with independent palettes are supported by the format.
|
|
#
¿
Jan 3, 2014 04:02
|
|
|
RobertKerans posted:Not terribly scientific, but it works and forces you to actually learn why things are there, and what depends on what: set up a folder with a test page (containing exactly what want to achieve in HTML form) + grunt with livereload and less plugins. Grab an uncompiled copy of the bootstrap less files, then add them bit by bit to your project folder, compiling to CSS as you go. The terminal will scream at you at first, and you'll be forced to either write substitutions or locate the dependency and understand why it's there. Not the greatest method in the world, but it does work. Agreed. The key concepts are "use the LESS source" and "don't pull the library's CSS in wholesale." I hear LESS isn't the ideal CSS preprocessor, but it works for what I'm doing right now and it's really the best way to work with Bootstrap. I haven't used grunt and just drop less.js into my page at first since it's easy to get started and easy to debug.
|
|
#
¿
Jan 6, 2014 18:44
|
|
|
pipes! posted:I figure you can get about two, "Oh hey, I noticed" in before you look like a total rear end and sour any future working relationship. Okay that's the second time I've seen this comment in the last page of posts. What are the pitfalls of absolute padding? When has it bitten you? Why only padding and not margin, border, width, and other size elements? I remember there being a big push towards away pixel-based measurements overall many years ago, but when I talked to CSS people at conferences lately concern seems to have waned. I've just been using pixels again lately since it seems to work just fine for my projects, but it'd be good to know when that's bad.
|
|
#
¿
Jan 9, 2014 23:19
|
|
|
Pollyanna posted:That's what I figured. But the URL looks bit funky (/stocks?symbol=IBM%2CGOOG%2CAAPL). Not a big deal to me, but is that discouraged? You could always use a URL rewriter or routing module or something to make urls that look like /stocks/IBM-GOOG-AAPL if you wanted instead of using a query string. Dashes don't normally get escaped so it'd appear nice.
|
|
#
¿
Jan 12, 2014 02:13
|
|
|
Kobayashi posted:I'm designing a SPA with iPad-inspired popovers instead of modal dialogs; can anyone recommend a good library? I'm looking at qTip right now, but I thought I'd see if there's anything better. If you're looking for something that looks similar, the Popovers in Bootstrap look pretty close. The default triggering/disposing logic sucks for what I've tried to use it for, but if you're replacing a modal with this it might work well for you.
|
|
#
¿
Jan 14, 2014 15:51
|
|
|
I can't stand this obsession with blanket statements and pure dogma in software engineering. It's like people want to declare rules and be the guy with the blog about that or something. It's totally about context. On a link aggregator, you better believe I want to always open in a new tab so I don't lose my place.
|
|
#
¿
Jan 18, 2014 02:54
|
|
|
RobertKerans posted:Yesterday at work I got sent the link to the css-tricks article discussing this, and aggregators were the first thing I said in response to it. But then the first response I got was 'well, yeah, but back button'. I'd rather have shitloads of tabs open (and this applies to mobile as well), and read through them at leisure, but another, say 50%, of people go open, read, back. I've always gone blank for external, but it's absolutely subjective. Well, it doesn't matter about a back button anyway if the list implements infinite scrolling. Sure, you could hash nav on the page they're on and fast-forward to that on back navigation, but that's pretty inefficient compared to just opening in a new tab by default. You could still implement the other behavior too - you're just saving yourself some bandwidth using target="_blank".
|
|
#
¿
Jan 19, 2014 03:48
|
|
|
Kobayashi posted:My impression is that grunt is essentially make for the web and Bower is a good idea whose time is not quite here yet. Last time I tried Bower, I liked the idea of pulling in dependencies without seeking them out and downloading them manually, but I found that the packages were inconsistently named, with inconsistent directory structures and variable amounts of extra cruft inside, so I'd still have to go through and move things around / delete leftovers. Maybe it's gotten better since then, but at the time, it wasn't much better than manually downloading the files I needed. I had this impression of NuGet (Microsoft's package manager) when I used it the first time. I'm pretty sure it'd be true of any package manager that tries hard. In MSFT space though, everyone likes Content/Scripts folders for some reason so that kind of works a little. If you're in an ecosystem where there isn't a canonical framework for X then I can see how that'd be an even bigger problem. Just think about the process of bulk-updating your dependencies though. It's way too easy and naive. If you update a dependency, you have to test the change anyway. That's a lot more work than just downloading it and sticking it in there. For that reason I'm starting to prefer downloading from the package manager since it's easy and centralized when I'm starting a project, then arranging things the way I want and deleting the cruft. Then, I never use the package manager again unless I want a completely new component. It's okay that I can't update it through the package manager, since I won't be doing that lightheartedly anyway. However, I do get whatever extra bootstrapping the package manager gives me in terms of config file updates or whatever you need done. (this might just be a NuGet thing) So, the conclusion I'm coming to as I write this post is that package managers are lovely app stores that you can't effectively update from. I'd love someone to show me the light if this is wrong. glompix fucked around with this message at 17:41 on Jan 21, 2014 |
|
#
¿
Jan 21, 2014 17:35
|
|
|
Bognar posted:If you don't deviate from a lot of the conventions that Microsoft sets up in its frameworks, NuGet is wonderful. Also, if you installed a package initially through NuGet, it's so ridiculously easy to update it through NuGet that I don't understand why you would do something else. Though in my opinion, the best part about it is the Visual Studio integration for package auto-restore. You can pull down a copy of code with a packages.config file and no libraries, open it in Visual Studio, and try to build then VS will automatically grab all the dependencies before starting the build. This argument has been beat to death on these forums, but I think it's a better idea to package your binary dependencies with your code. Your software may need very long-term support, (the project I'm working on right now has been around 10 years and just will not die) and so it makes sense to not solely rely on external sources for that kind of thing. But yeah NuGet isn't so bad because of those conventions. "Update all" just isn't something you're going to ever do in a big project with many little front-end dependencies, so it kind of defeats some of the advantage of having the system past the initial bootstrapping. I think I'm splitting hairs at this point though. NuGet and Bower are great overall.
|
|
#
¿
Jan 22, 2014 15:42
|
|
|
Chat suggestion: just try socket.io or signalr. It's pretty drat simple these days.
|
|
#
¿
Jan 24, 2014 06:13
|
|
|
pokeyman posted:Why not? Is there some standard that says this? XHTML 1.0, HTML 4 for sure. In HTML5, you can wrap block elements in an anchor only. Conceptually the rule makes sense. Block level elements have different layout rules which collide with inline elements. (is it page width or auto width? it's undefined) I'd stick with the rules. It's just a safer bet with such a diverse browser ecosystem. I guess you can just paper over those differences with CSS though, but it's still undefined behavior. glompix fucked around with this message at 22:39 on Feb 7, 2014 |
|
#
¿
Feb 7, 2014 22:36
|
|
|
pokeyman posted:Finally, is that undefined behaviour? I recall part of HTML 5 was a push to standardize error handling. This certainly happened with parsing, but I can't find much about content model violations in the spec. Well, maybe it is defined as an error, but when would you ever want to write erroneous markup? I used to worry about this kind of poo poo in 2008-2009 and it's really not worth it to obsess over, but it's just good to know where the boundaries of the spec are even if you choose to ignore them for whatever reason.
|
|
#
¿
Feb 7, 2014 23:12
|
|
|
cbirdsong posted:iosSlider has really nice touch support: https://iosscripts.com/iosslider/ I've never seen a single component with such a detailed licensing structure. Looks solid though.
|
|
#
¿
Feb 21, 2014 18:19
|
|
|
I do really wish there was a Javascript interpreter for Sass like LESS has. That'd make diving in quicker. If you in Windows, check out WebMatrix. Getting Sass set up in visual studio is pretty easy so this may be too. http://www.hanselman.com/blog/WebMatrix2FrontEndWebDevelopersTakeNoteASPNETPHPNodejsAndMore.aspx
|
|
#
¿
Feb 22, 2014 17:49
|
|
|
kedo posted:I have an idea for one of my upcoming work projects which would require giving users the ability to draw simple shapes, and then performing some sort of basic pattern analysis on the shape. Like, user draws something that looks like an L and some JS examines it and says, "oh, that's probably an L." This might be overkill, but you could always code your own OCR. There's an example in "Machine Learning in Action" that I'm sure you could port to Javascript. You'd definitely need a good training set though. I say this because the one you linked kind of sucks at recognizing my mouse hand-writing which is admittedly bad but not that bad. You also said you might want to do shapes and not just handwriting. 
glompix fucked around with this message at 21:18 on Feb 28, 2014 |
|
#
¿
Feb 28, 2014 21:15
|
|
|
I don't really care about the way Atom looks or behaves by default. If it's like emacs, it's about how you build your own editor. This is really exciting to me. About to try it out. edit: poo poo, I don't have a Mac. At least I'm starting a new job in 4 weeks and will get one.
|
|
#
¿
Mar 1, 2014 15:17
|
|
|
|
| # ¿ May 8, 2024 21:54 |
|
|
That sucks, man! You could always run a local git repository and just never put it anywhere except your machine. There's still lots of value in that for change tracking.
|
|
#
¿
Mar 2, 2014 16:49
|
|
