EPTOr do you just do the following?
Company B
- Experience
Company A (now called Company B)
- Experience
The Big Programming Thread - Page 954
| Forum Index > General Forum |
Thread Rules 1. This is not a "do my homework for me" thread. If you have specific questions, ask, but don't post an assignment or homework problem and expect an exact solution. 2. No recruiting for your cockamamie projects (you won't replace facebook with 3 dudes you found on the internet and $20) 3. If you can't articulate why a language is bad, don't start slinging shit about it. Just remember that nothing is worse than making CSS IE6 compatible. 4. Use [code] tags to format code blocks. | ||
|
sc-darkness
856 Posts
Or do you just do the following? Company B - Experience Company A (now called Company B) - Experience | ||
|
ShoCkeyy
7815 Posts
| ||
|
WolfintheSheep
Canada14127 Posts
| ||
|
bo1b
Australia12814 Posts
| ||
|
berated-
United States1134 Posts
On April 11 2018 12:55 WarSame wrote: I feel like I'm missing something with Docker. What is the ideal development to production pipeline supposed to look like? What is Docker's place in there? Do you need 1 docker-compose per environment? How do you handle environment changes between containers? When I am developing(say a Flask app) I have a mounted volume to auto update my code. When I am done developing I commit to my own branch("flask-101013033", maybe). QA comes in and verifies that this branch works correctly by pulling the branch, building an image from the branch and making a container from the image. We then move it into integration testing. I merge my code into the integration branch. We then build another image/container for this branch which QA tests for integration. Finally, we move this image into production. Is that a reasonable approach? How do Kubernetes and Docker Swarm come into play? Ideal is a bit weird, as deployment is so culture related that it's hard to say... Our company is using it as this.. I don't use docker in my development environment except to run databases. It's too much of a pain to try to get in and out of the container or mapping mounts for my liking. However, running the database in a docker container so it can be local is fantastic (especially considering they have a sql server on linux image). If you had a large toolchain that is needed for a dev environment I can see how it's attractive to want to use docker for dev though. One of the strongest selling points of docker is that it gives you the ability to move the exact container between different environments. The container you built in your ci loop, is the container that goes to dev, qa, prod. Not that your setup doesn't work with rebuilding the container, but, trying to extract your configuration out of the build and then pushing the container through the deployment phases just means you're probably more likely to make sure that what you tested in the environment was correct. This is used pretty often with CI/CD. If you're using deploy branches and actually rebuilding you may not get as much advantage at of this. We also use docker as a part of our gitlab ci loop. When the gitlab pipeline kicks off it spins up a docker builder that has our tools on it to build our applications. It can link in databases, or even selenium servers to run your automated tests against, or any external service that you might have depended on a vm or dedicated server for. Kubernetes and swarm come in for orchestration -- if you are using microservices and need to get each of the different containers communicating with each other, these tools make it a lot easier to pop up a set of microservices that comprise a full application. Our main website could end up using anywhere from 6-10 services that make up the one app. Trying to "deploy" the entire thing manually is a lot of effort where as those tools make the whole thing a lot easier. Kubernetes can also do a ton of stuff with controlled rollouts and versioning and ingress controllers and a whole bunch of stuff that I don't even fully understand... Docker is really cool, but, the entire community is still pretty young and things are changing very rapidly. It's definitely buzzwordish but unless you are deploying quite a few application instances it's a lot of overhead for not a ton of gain... except for running your databases in dev on your machine, that's a life saver. | ||
|
Neshapotamus
United States163 Posts
On April 12 2018 07:55 bo1b wrote: I finished reading Haskell programming from first principles, anyone remotely interested in functional programming should consider it. A genuinely great book. This man gets it! I am also reading this book, but only at chapter 18. | ||
|
Manit0u
Poland17706 Posts
On April 12 2018 03:31 Bog wrote: Some ideas; - Check is other transactions aquire locks on your data. Lock monitoring. - Check for any triggers that might be hit on this transaction. SELECT * FROM information_schema.triggers - Check the explain plan for anomalities. EXPLAIN ANALYZE UPDATE "table_name" SET "last_seen_at" = now(), "updated_at" = now() WHERE "table_name"."id" = "uuid";(make sure equality predicate is correctly applied on the primary key. Equality predicates with different types, e.g. x::uuid = y::varchar can sometimes cause unexpected behaviour) - Check your isolation level SELECT current_setting('transaction_isolation');. Concurrent access to data could be severely hindered when your isolation level is unnecessarily high (e.g. 'serializable'). Consider setting the level to 'read committed' if your business logic allows.- When using JDBC, check if your transaction is committed at the earliest moment your business logic allows. Extra care should be taken when defaultAutoCommit=false for your JDBC connection or connection-pool property, then always close the transaction/connection explicitly in your code. Transaction isolation is set to 'read commited'. I think the problem might be caused by exceptionally high server load at certain times (we can have up to 500 active db connections) since query planner looks good, there are no locks or triggers on that either. Will dig deeper, created a script that dumps system stats when such a thing occurs. | ||
|
sc-darkness
856 Posts
The sad thing is I'm sure there's a lot more to Git than I currently know. | ||
|
bo1b
Australia12814 Posts
On April 13 2018 02:29 sc-darkness wrote: While it's useful I learnt more about Git these last few days, I feel like my mind was "raped". Going from merge-commits, cherry-pick & logs to (interactive) rebase, squash, using stashes and fixing existing commits is quite a jump. I'm not a dumbass, but if a guy with CS degree has a difficulty, then obviously this tool isn't designed to be easy to use. It's powerful, I admit that. The sad thing is I'm sure there's a lot more to Git than I currently know. Honestly git falls into a category with a few other tools that I honestly wish were covered to some degree at university, proper terminal usage comes to mind as an example. | ||
|
Excludos
Norway8243 Posts
On April 13 2018 02:35 bo1b wrote: Honestly git falls into a category with a few other tools that I honestly wish were covered to some degree at university, proper terminal usage comes to mind as an example. It wasn't directly covered in my degree, but we were told to use git for most of our projects. One of the courses I took in Australia also covered git in detail, even if it really had nothing to do with the subject itself. I think it's just some of the teachers realizing students need this and just hammered it where they could. | ||
|
Blitzkrieg0
United States13132 Posts
On April 13 2018 02:35 bo1b wrote: Honestly git falls into a category with a few other tools that I honestly wish were covered to some degree at university, proper terminal usage comes to mind as an example. The problem with covering tools is that you can't teach them all. Are you working in a windows or linux shop? Are you using git, SVN, clearcase, some other version control? I understand the sentiment, but it isn't practical to learn everything. | ||
|
Excludos
Norway8243 Posts
On April 13 2018 03:44 Blitzkrieg0 wrote: The problem with covering tools is that you can't teach them all. Are you working in a windows or linux shop? Are you using git, SVN, clearcase, some other version control? I understand the sentiment, but it isn't practical to learn everything. Sure, sure. But you should learn a version control tool to have some foundation to later learn other's from. And since git is the mother of them all, that's the one you should be learning (Just like, in my not super humble opinion, you should be learning C++, or at worst Java, when learning to program (And not a scripting language like Python or JS..or Matlab), so you'll have a good foundation to learn other's from. I think most CS courses probably do this, but a lot of the other engineering courses like cybernetics or electrical is not equally good here. I have met fully educated electrical engineers who has never touched C... good luck there buddy. You're going to have a fun time with micro chips) | ||
|
Manit0u
Poland17706 Posts
On April 13 2018 02:35 bo1b wrote: Honestly git falls into a category with a few other tools that I honestly wish were covered to some degree at university, proper terminal usage comes to mind as an example. You mean, doing stuff like that?
Console is amazing. Everyone should learn to use it and love it  | ||
|
bo1b
Australia12814 Posts
On April 13 2018 04:02 Excludos wrote: Sure, sure. But you should learn a version control tool to have some foundation to later learn other's from. And since git is the mother of them all, that's the one you should be learning (Just like, in my not super humble opinion, you should be learning C++, or at worst Java, when learning to program (And not a scripting language like Python or JS..or Matlab), so you'll have a good foundation to learn other's from. I think most CS courses probably do this, but a lot of the other engineering courses like cybernetics or electrical is not equally good here. I have met fully educated electrical engineers who has never touched C... good luck there buddy. You're going to have a fun time with micro chips) Having just taught myself a good amount of functional programming I wish I started with haskell, then went to c then to c++ and then nothing but theory. I feel like when you understand functional programming you understand programming a lot better, at least I did. Past learning c++ to a competent level you honestly shouldn't need any hand holding at all to learn another language. And I believe that theory side of things, and the math side of things should be heavily expanded upon. On April 13 2018 03:44 Blitzkrieg0 wrote: The problem with covering tools is that you can't teach them all. Are you working in a windows or linux shop? Are you using git, SVN, clearcase, some other version control? I understand the sentiment, but it isn't practical to learn everything. It's not practical to learn everything but every single graduate with a degree in computer science should have done some work into operating systems, at which point the two main varieties are microkernel and monolithic, and the two main examples of those are unix and windows. There is a cross over between computer science and a trade, so it only makes sense to learn some level of practical application for those two things, powershell and terminal don't take infinite time to learn, and git mercurial and svn don't either. | ||
|
Deleted User 3420
24492 Posts
| ||
|
bo1b
Australia12814 Posts
| ||
|
sc-darkness
856 Posts
On April 13 2018 06:08 travis wrote: I had to use git for one class, but honestly I don't get it. It just seems like yet one more thing with a ton of annoying shit to learn. Why is it so important to learn all that stuff exactly? Why do I need to know anything beyond "git push" ? Because having the following commit messages is annoying (I did it in the past too): Commit 1: Fixed bug with localisation. Commit 2: Ok, fixed bug with localisation this time. Commit 3: Ok, finally fixed the bug!!! Instead, you could squash them and end up with: Commit 1: Fixed bug with localisation. Simple and short. You're also able to edit commit messages if you think they weren't initially useful. Another useful feature is if you do auto formatting (e.g. IDE does it) in several commits, you could move it all into one commit, then the other commits could be cleaner (no auto formatting). It's much easier for reviewers of your code to focus on actual changes. | ||
|
bo1b
Australia12814 Posts
And anyone who uses emacs knows about magit, which is possibly the best version control utility the world has ever seen. | ||
tofucake
Hyrule19198 Posts
| ||
|
WolfintheSheep
Canada14127 Posts
On April 13 2018 06:08 travis wrote: I had to use git for one class, but honestly I don't get it. It just seems like yet one more thing with a ton of annoying shit to learn. Why is it so important to learn all that stuff exactly? Why do I need to know anything beyond "git push" ? Seems like the standard just-out-of-Uni attitude. 1) I'm smart, I'll just make things that work. 2) This looks terrible. Oh well, I can just change this. 3) Hmm, why didn't that do anything. Oh, I see, I need to change these too. 4) Oh fuck what did I just do panic panic boss is going to find out soon. 5) Why isn't there a repo?! | ||
| ||
