|
PT6A posted:My other general rule is that computers/browsers (with a few exceptions) are pretty goddamn overpowered for displaying your average website, so if there's a choice between a method that will be easy for you to maintain, and a method that will be slightly more efficient for the computer, it's best to make your life easier and let the browser pick up the slack. Sounds like a bad approach for mobile network considerations?
|
#
?
Nov 11, 2018 15:04
|
|
|
|
| # ? Jun 9, 2024 00:26 |
|
|
The Dave posted:Sounds like a bad approach for mobile network considerations? I'd argue the opposite. Where mobile network bandwidth and latency is a primary concern, it's best to use the local resources of the smartphone and do as much as possible client-side, rather than worrying about how efficient the page is on the browser itself.
|
|
#
?
Nov 11, 2018 15:41
|
|
|
22 Eargesplitten posted:Okay, thanks. Currently not using any preprocessors, that's one of the many things I need to learn to do after I get this first one done. Be a cool kid and check out Stylus, it's the most fun preprocessor there is. I thought the syntax was dumb and gimmicky at first but I actually prefer it over Sass now. What's the deal with TJ Hollowaychuk anyway, I think he's a pseudonym for a group of devs making cool stuff 
|
|
#
?
Nov 12, 2018 22:42
|
|
|
Hey guys I suck at html, I'm trying to use flexbox to display a row of 5 items of varying width, and I want them all to have equal width that matches the width of whatever content is widest. And also centered so like: -------------------------------[...............]-[...............]-[...............]-[...............]-[...............]------------------------------- Imagine they all have some text inside. I tried this but it doesn't seem to grow to the biggest width: https://codepen.io/anon/pen/rQWbeX Can someone point me in the right direction?
|
|
#
?
Nov 13, 2018 03:33
|
|
|
FateFree posted:Hey guys I suck at html, I'm trying to use flexbox to display a row of 5 items of varying width, and I want them all to have equal width that matches the width of whatever content is widest. And also centered so like: I will post code tomorrow if nobody else has (on phone and going to bed ), but look up CSS-Tricks visual guide to flexbox. Should be easy for you to figure out from that, and it's handy to have bookmarked.
|
|
#
?
Nov 13, 2018 05:17
|
|
|
Lumpy posted:I will post code tomorrow if nobody else has (on phone and going to bed ), but look up CSS-Tricks visual guide to flexbox. Should be easy for you to figure out from that, and it's handy to have bookmarked. I'm not sure it can't be done, but it certainly doesn't seem easy to figure out.
|
|
#
?
Nov 13, 2018 08:58
|
|
|
rohan posted:Unless I'm misunderstanding the question, it's actually really tricky and I'm not sure even possible with CSS alone. Somehow you need to find the item with the widest content, and use the width of that item for every other sibling. Oh yeah.. I missed the "of the widest content" bit. That's what I get for reading the forums past my bedtime. If you want "equal widths", which is what I thought the question was, you can use flex: 1 on the items. Forcing them to be the width of one item's content is, like you said, not easy. But still bookmark this guide: https://css-tricks.com/snippets/css/a-guide-to-flexbox/
|
|
#
?
Nov 13, 2018 14:51
|
|
|
Could you pass the content width alongside the content itself?
|
|
#
?
Nov 13, 2018 15:51
|
|
|
Yeah, I'm not sure you can do that with CSS alone if your content is text. Text width is endlessly variable, and unless you use some JS to find the width of the widest item I'm not sure how you can accomplish that.
|
|
#
?
Nov 13, 2018 16:36
|
|
|
Is this something that html tables handle natively? somehow make all the cells the same width but not make the table 100% wide? If this is too obscure maybe I'll just find some other hack..
|
|
#
?
Nov 13, 2018 17:17
|
|
|
FateFree posted:Is this something that html tables handle natively? somehow make all the cells the same width but not make the table 100% wide? If this is too obscure maybe I'll just find some other hack.. No. Table cells size to content by default, so your cells would all be different widths, just like default flexbox items. If you want to size explicitly to arbitrary content size, I think you'll need JS.
|
|
#
?
Nov 13, 2018 17:25
|
|
|
Back-end question: I remember reading a while back in a blog I found via mailing list that sort of gave a top level overview of back-end stuff. One of the main points the author made was that the best way to make your back-end scalable is to make everything with a minimum of two database servers even if it's not necessary because it's easier to scale upward from that than it is to take the more trivial approach where you just have one back-end server that queries and serves up everything on its own. I might be slight misremembering that, though. My question is: where can I learn more about how to setup my own portfolio projects that way so that I can tell these recruiters that I know how to write a scalable web app even though I don't have years of professional experience doing it already?
|
|
#
?
Nov 14, 2018 09:55
|
|
|
Love Stole the Day posted:Back-end question: I remember reading a while back in a blog I found via mailing list that sort of gave a top level overview of back-end stuff. One of the main points the author made was that the best way to make your back-end scalable is to make everything with a minimum of two database servers even if it's not necessary because it's easier to scale upward from that than it is to take the more trivial approach where you just have one back-end server that queries and serves up everything on its own. I might be slight misremembering that, though. Depending on the technology, clustering database servers is primarily done for high availability purposes, rather than increasing scale-ability - that is, if you set up a cluster for say, SQL server, you're going to get lower performance than the equivalent on a single database. There are database technologies that can horizontally scale - some NoSQL solutions like cassandra will scale horizontally, and you can play tricks like using redis etc. as a cache to lower the load on a relational database if it's going to scale. Load balancing is more about improving scalability - you can have multiple application servers point to the same back end, and have a load balancer in front doing round robin - the point would be to lower CPU usage of the app servers, and the app servers can use distributed caching to lower the load. I just read about technologies, download the software, set it up, write performance tests and see what happens. TBH I would shoot for getting a front end job and learning this stuff at work while at the front end job rather than trying to go from no job -> back end, downloading random database technologies and writing code to test performance looks astonishingly like real work to a software development house, and you'll get user stories where you're expected to put a button that shows up on a web-page and a week of time to do it.
|
|
#
?
Nov 14, 2018 12:39
|
|
|
Love Stole the Day posted:Back-end question: I remember reading a while back in a blog I found via mailing list that sort of gave a top level overview of back-end stuff. One of the main points the author made was that the best way to make your back-end scalable is to make everything with a minimum of two database servers even if it's not necessary because it's easier to scale upward from that than it is to take the more trivial approach where you just have one back-end server that queries and serves up everything on its own. I might be slight misremembering that, though. I dunno about how to set something up that would look impressive in a portfolio, but in a real app, you probably shouldn't think about database scaling at all. It's a form of premature optimization, and it's optimizing something that almost nobody needs to optimize today. You can get enterprise-grade infrastructure such as an absolutely monstrous postgresql instance with hundreds of gigabytes of RAM that will scale to tens of thousands of simultaneous users at the cloud provider of your choice for almost nothing these days (well, almost nothing compared to what the engineering hours needed to architect some complicated distrubuted system). For 99% of web apps, scaling is never going to be an issue - the problem is getting someone to care. Big data needs to be absolutely monstrously big these days to really need distrubuted solutions. Code is horrifyingly expensive, not only to write, but also to maintain. It's very often cheaper in the long run to attack optimization problems with commodity infrastructure than with software engineering.
|
|
#
?
Nov 14, 2018 12:40
|
|
|
TheFluff posted:but in a real app, you probably shouldn't think about database scaling at all. It's a form of premature optimization, and it's optimizing something that almost nobody needs to optimize today. I don't disagree with you, and if I were asking because I wanted to create my own web app then I'd absolutely do what you recommend. The only way to gain experience performing these sorts of things when you don't already have a job doing them in the first place is to intentionally do premature optimizations like this simply for the sake of doing them because you'll be trying to interview with already established companies where their apps are already using those techniques. Otherwise, you can't say in an interview that you have experience implementing scalable web apps or scalable back-end infrastructures, and therefore you won't be able to solve that sort of problem if they test you on it in a technical interview, which in turn makes you less employable. Bruegels Fuckbooks posted:I just read about technologies, download the software, set it up, write performance tests and see what happens. TBH I would shoot for getting a front end job and learning this stuff at work while at the front end job rather than trying to go from no job -> back end, downloading random database technologies and writing code to test performance looks astonishingly like real work to a software development house, and you'll get user stories where you're expected to put a button that shows up on a web-page and a week of time to do it. Again, I don't disagree. I've been working on getting a front-end job for a long time now. It's just that recent feedback I've received tells me that if I want to make myself more valuable for a job, then I should focus on making the employer feel like they could put me anywhere in the stack, which I can only do if I know enough back-end skills and techniques to be able to do this sort of thing, which in turn means that I need to pivot from front-end portfolio stuff more toward back-end portfolio stuff now. When I look at back-end job openings right now I see requirements like __ years of experience with Java Spring or some other large scale distributed systems technology, and therefore that's what needs to be in my portfolio.
|
|
#
?
Nov 14, 2018 13:38
|
|
|
Basically if you're on MySQL use Galera for failover and you'll get read replicas for free. Postgres has its own good docs about HA setups and read replicas that tell you pretty much exactly what to do.
|
|
#
?
Nov 14, 2018 15:01
|
|
|
Another approach would be to break up your backend into micro-services. There was a talk from a Netflix engineer who went through the process, but it does come with a lot of challenges, such as keeping api versions aligned. The general idea is that each part of the application can be scaled separately to meet emerging demands. edit: https://www.youtube.com/watch?v=CZ3wIuvmHeM bigmandan fucked around with this message at 15:38 on Nov 14, 2018 |
|
#
?
Nov 14, 2018 15:22
|
|
|
If you're trying to get a job at a place that's hiring at a junior/entry level and they want their candidates to know how to set up a distributed cloud-based system with load balancing and failover, their requirements are insane, and their whole environment likely is, too.
|
|
#
?
Nov 14, 2018 16:27
|
|
|
Munkeymon posted:If you're trying to get a job at a place that's hiring at a junior/entry level and they want their candidates to know how to set up a distributed cloud-based system with load balancing and failover, their requirements are insane, and their whole environment likely is, too. You have a good point. It sounds like what I'm asking about here is too advanced of a skill to demonstrate in a project. I think I just need to go back and figure out something else back-end related to try and demonstrate in a project. Nice link! I will watch this
|
|
#
?
Nov 15, 2018 16:37
|
|
|
Love Stole the Day posted:You have a good point. Another cool thing is they open sourced some of the code to create their infrastructure. http://netflix.github.io/
|
|
#
?
Nov 15, 2018 16:40
|
|
|
I have a demo project I'm trying to modify; right now it shows connection info for the server we're currently using, so we can dramatically demonstrate that "At the push of a button we've just migrated this data to the other database, and also switched our app to use the other DB". Right now, it's SQL-only, and I'm supposed to make it work with Mongo/Cosmos. Right now, this data comes from the server calling "SELECT VERSION();" on the DB and reading the result, what's the equivalent in Mongo? I've found db.version, but I don't know how to call that from an outside URL, especially using the C# driver.
|
|
#
?
Nov 20, 2018 00:42
|
|
|
My sales guys are freaking out because they discovered GTMetrix and YSlow and are scared other agencies are going to swoop in and steal their clients by showing their clients a bad YSlow score. A large part of the negative scores seems to be the Google Maps embed. They want a Google Map on the homepage of the site, but it seems to be drawing a lot of resources, e.g. HTTP requests, external stylesheets, external JS, etc. Taking away the Google Maps embed reduces the load time from 3.2s to 1.5s alone! Is there any way to improve this? YSlow specifically is hitting hard on "Use a Content Delivery Network (CDN)" - but that seems incredibly overkill if it's just some simple images that aren't on any other site also, right? Like should I be setting up a CDN for every single site that's going to have maybe a hundred visitors a month? They host on Media Temple, which appears to be in Virginia and/or California, and we deal with mostly small east coast companies. I don't know much about CDNs, but I thought they were for major things that are the same across hundreds of thousands of sites such as jQuery (which I don't use anymore) or Google Fonts. And then there's "There are 10 static components without a far-future expiration date." - which are mostly things like one Google Fonts entry and some random stuff from a third party tracking plugin. There's no way to set an expiration date for those, right? Also, while I'm typing this and remember - a Contact Us page is still important, right? I know it's the first thing I look for when I go to a website, but some sales people don't want me to include one and just stick a contact form on the bottom of the site. Anything I can see to show how much a Contact Us page captures leads, in general?
|
|
#
?
Nov 20, 2018 06:16
|
|
|
Hey thread, is there any consensus or discussion on honeypots/recapcha tactics? We've started picking up more and more spam over the last few months and are looking at alternative solutions to our honeypot method which has served us well for over a decade... even the google recapcha /nocapcha seems to be failing us and so I'm currently thinking how can we build something in that bulletproofs our forms once more. Surely others are running into this problem so hopefully there's a method which can prevent these stupid form entries from bots?
|
|
#
?
Nov 20, 2018 11:35
|
|
|
LifeLynx posted:And then there's "There are 10 static components without a far-future expiration date." - which are mostly things like one Google Fonts entry and some random stuff from a third party tracking plugin. There's no way to set an expiration date for those, right? I don't think so. Something I tried, for google analytics, is to cache localy the ga.js file. But is annoying to have something this fragile that can stop working one day, just because Google ask me to cache in X way, and google itself don't cache in X way.
|
|
#
?
Nov 20, 2018 14:26
|
|
|
LifeLynx posted:My sales guys are freaking out because they discovered GTMetrix and YSlow and are scared other agencies are going to swoop in and steal their clients by showing their clients a bad YSlow score. Load the page, then embed the map. Put in a placeholder (empty DIV or get fancy and make a very simplified static image of the map) and after the page is done loading, use JS to embed the map in that spot.
|
|
#
?
Nov 20, 2018 15:01
|
|
|
Another year passed, another year feeling like literally the only architect decision I've made is a bad one (using Angular) Though, for Angular, React and Vue the (second) best and (second) worst thing about them is their programming style.
|
|
#
?
Nov 20, 2018 16:51
|
|
|
The Merkinman posted:Another year passed, another year feeling like literally the only architect decision I've made is a bad one (using Angular) Using something popular and with good documentation is never a bad decision. At least you are in a good position when you jump to something better. Think where you would be if you where building something out of a dailywtf-com article, like a custom in-house javascript-like programming language.
|
|
#
?
Nov 20, 2018 17:02
|
|
|
Lumpy posted:Load the page, then embed the map. Put in a placeholder (empty DIV or get fancy and make a very simplified static image of the map) and after the page is done loading, use JS to embed the map in that spot. What's the best way to do that? I tried: code:
|
|
#
?
Nov 20, 2018 17:04
|
|
|
LifeLynx posted:My sales guys are freaking out because they discovered GTMetrix and YSlow and are scared other agencies are going to swoop in and steal their clients by showing their clients a bad YSlow score. Those speed metrics are all "best case" scenario and should be taken with a grain of salt. Also GT et al give weird weighting to certain factors like "oh use a CDN" when it affects less than 10% of the images or the like. Same with unexpired resources, especially given you don't have control over a lot of them. I will say having a Google Map on the landing page might not be the best idea but can be mitigated as Lumpy notes, but I'd consider moving it off entirely to a different page (like maybe, a Contact Us page where it makes sense?). I generally pay more attention to the Google metrics regardless since that theoretically has more ranking impact if they ever get that off the ground. EDIT-What I've found more important is the actual amount of bandwidth going down the pipe, and how quickly it is marshalled. The waterfall diagram helps with that. Scaramouche fucked around with this message at 19:33 on Nov 20, 2018 |
|
#
?
Nov 20, 2018 18:36
|
|
|
The Merkinman posted:Another year passed, another year feeling like literally the only architect decision I've made is a bad one (using Angular) https://2018.stateofjs.com/opinions/opinion-building_js_apps_overly_complex the number of people who don't think it's too complex to build a JS app is growing 
|
|
#
?
Nov 20, 2018 21:37
|
|
|
It seems to me the complexity in building JS apps is growing as necessary to accommodate the expectations of users. The barrier for entry is higher, but far more can be done with it and far more easily than five or ten years ago. When all we were doing was accordions and real basic ajax work the barrier for entry was understandably lower. Increasing complexity seems inevitable to me. There must be some law that states the longer a technology is used the more complex it becomes.
|
|
#
?
Nov 20, 2018 22:52
|
|
|
LifeLynx posted:My sales guys are freaking out because they discovered GTMetrix and YSlow and are scared other agencies are going to swoop in and steal their clients by showing their clients a bad YSlow score. Tell your sales guys all their lovely 3rd party analytics plugins are what is killing their site speed. Hell even loving GTM adds 2 seconds onto my time to interactive. You can set an expiration date by hosting a copy of the resources locally and setting the expiration there. It's a dead end though, it didn't improve my lighthouse score by even a point. And if you invest a little bit of time to getting a good S3/Cloudfront build going, you wont have to worry about CDNs. It's basically magic. Raskolnikov2089 fucked around with this message at 03:49 on Nov 21, 2018 |
|
#
?
Nov 21, 2018 03:38
|
|
|
Host your own Matomo instance and disable all the privacy features imo
|
|
#
?
Nov 21, 2018 03:41
|
|
|
I've been working at a pretty large institutional website for almost 3 years as a designer. I was brought on when they had 2 months to launch the forward-facing website with a ton of content and had no design and obviously no code. All they had was the decision to work in Drupal. They wanted something that wasn't a theme or anything visibly framework-based and custom-made. They hired an outside team to handle all development, too. Unsurprisingly, it's been a nightmare, adding a news portal and a bunch of other stuff where everything is "urgent" and needs to come out in 2 months, had predictably ended up as a buggy, inefficient mess that breaks under the slightest scrutiny, and completely unrelated parts of the site breaking every single deploy. The IT team has implemented tremendously cumbersome bureaucratic workflows which tend to mean about a month's time before anything can be deployed by now. I fought to get PatternLab implemented to have some documentation and have a more visible code base to target fixes against, but it's been kicked along to make room for more "visible" development (ie. new features/areas) for a year and a half so far. Our permissions are chimeric and have recently collapsed, performance is in the gutter, our css is full of !importants. I constantly bring up the fact that we need a few maintenance sprints, and am always told we need to get new feature "X" out the door before that. A new hire joked about throwing everything out and starting from scratch and I laughed saying it was the best thing we could do. Recently we hired a team for a quick side project for making quick websites for random stuff (like event pages and smaller project pages) and I'm getting frustrated. I'm more of a UI/UX guy and really enjoy designing tools and dashboards, and most of my time is now spent as QA writing bug reports and answering emails telling them all the fires are fine and not to worry, that they will be put out soon. I'd look for something else, but the job security and salary are pretty good and benefits are great. Still getting kind of burnt out though.
|
|
#
?
Nov 26, 2018 22:52
|
|
|
I'd say job security is an illusion, you could make more at a new job, and going home not burnt the hell out at the end of the day is a wonderful feeling.
|
|
#
?
Nov 26, 2018 23:56
|
|
|
I'd also highly recommend immigrating and then just working for a company in the US remote.
|
|
#
?
Nov 27, 2018 00:23
|
|
|
Love Stole the Day posted:Back-end question: I remember reading a while back in a blog I found via mailing list that sort of gave a top level overview of back-end stuff. One of the main points the author made was that the best way to make your back-end scalable is to make everything with a minimum of two database servers even if it's not necessary because it's easier to scale upward from that than it is to take the more trivial approach where you just have one back-end server that queries and serves up everything on its own. I might be slight misremembering that, though. I think really what you're asking for has been solved again and again and and yet again by the major infrastructure providers, and even by the more affordable providers too. If you're really determined, I would learn how to set up those technologies rather than trying to roll your own. ModeSix fucked around with this message at 00:49 on Nov 27, 2018 |
|
#
?
Nov 27, 2018 00:46
|
|
|
I've got a few essay questions on the pre-work I've mentioned before including stuff that I haven't seen covered anywhere. It's asking about how many instances of client-side assets are created, how many instances of server-side code are available, and how many instances of the dbs connected to the server apps are created. I haven't seen anything about any of that, is there anywhere with a good rundown on that type of thing that I could read? I don't want to brain dump / cheat on it, I just haven't found any good resources for me to learn about it. My instinct is to say one instance of the client-side assets per tab, but I also don't know if there's caching in modern browsers that would cut that down. I'd think server-side code availability would vary wildly depending on the server being used and whether it dynamically sizes or has a static number of instances assigned. Same thing with DB instances, especially because I don't know if it's talking about the browser's database (abbreviation IDB if I'm remembering right) or the web server's database.
|
|
#
?
Nov 29, 2018 20:14
|
|
|
Crossposting from the .NET thread:quote:Is there an application template somewhere I could use with the dotnet tooling that would give me : ASP.NET Core 2.1 + Semantic UI + TypeScript + yarn all ready? I realize I'm asking a lot as this is a very specific stack (although a popular one I reckon). But I'm jut a hobbyist and setting this up takes me way more time than it should.
|
|
#
?
Dec 4, 2018 18:29
|
|
|
|
| # ? Jun 9, 2024 00:26 |
|
|
OK, my main site is OctoberCMS (Laravel based CMS) on ([https://www.[domain].com) and I want to feature hosted websites and web apps as subdomains to that (https://www.[domain].com/live/[example]) however, the routing on OctoberCMS is assuming anything other than index is asking for something on OctoberCMS and is throwing up a 404. How do I get the routing to ignore the live folder in OctoberCMS (Or, failing that, Laravel?) Edit: Nevermind, I forgot how subrouting works, I am a goit. ConanThe3rd fucked around with this message at 09:16 on Dec 12, 2018 |
|
#
?
Dec 12, 2018 07:49
|
|
