Obviously selecting the right buildpack is critical in order to make secure, performant, and resilient applications. Heroku seems to work with the dev community to create and share buildpacks. A side effect is that there may not be a single, 'obvious' choice for a given tech stack.
Random example (R)
If we wish to use an R buildpack, a quick google show this popular buildpack is out of date and no longer supported, but we can easily find these three (just examples) here, here, and here
Another random example (cairo)
The cairo buildpack - googling shows numerous in the top results, it's not clear which to use out of, say, the top four on google here, here, here (this one clearly states deprecated, so we can rule that out), and here
Question
What are the main criteria against which we should assess heroku buildpack when faced with >1 that look like they will do the job?
I think the answer here depends on how much you're willing to trust someone else's code.
None of the buildpacks you mentioned are officially supported by Heroku. They are community buildpacks, maintained by the community, and they implement the Heroku Buildpack API.
The Buildpack API is open for anyone to implement. So the most discerning users will write their own buildpacks. Otherwise, you'll have to decide which community you trust the most.
Related
I'm trying to successfully submit my Electron app to the Mac Store. I want to put it into TestFlight, similar to how I am submitting the corresponding iOS app. Ultimately, I want both of these processes to be part of a CI pipeline. The iOS version of this is ready to go.
For the Electron app, I'm using Electron-Builder to do the packaging and signing, and leveraging electron-builder-notarize to get past the Apple notarization step. After about a week of diving down several rabbit holes that became full-fledged rabbit warren metropolises, I finally have a working combination of configuration settings, certificates, plists, and pixie dust in place that my pipeline successfully builds and notarizes my "mas" target package.
I thought that once I had gotten past the rather hellish notarization step, I would be pretty much out of the woods -- after all, Apple 'notarized' it (and sent me an email to prove it). But when I use the Apple Transporter app to post it to the App Store (I'm using this manually for now until I prove this works, then later I'll use the CLI commands for it to complete my CI implementation) it reports the app is rejected with 5 issues. These are actually variations of the same thing:
Electron.app' is already in use by another application.
Electron.app/Contents/Frameworks/Electron Helper (GPU).app' is already in use by another application.
/Electron.app/Contents/Frameworks/Electron Helper (Plugin).app' is already in use by another application.
Electron.app/Contents/Frameworks/Electron Helper (Renderer).app' is already in use by another application.
Electron.app/Contents/Frameworks/Electron Helper.app' is already in use by another application.
I thought Electron-Builder was supposed to deal with these internal ids. My "mas" block of the config looks like
"mas": {
"type": "distribution",
"hardenedRuntime": false,
"provisioningProfile": "embedded.provisionprofile",
"entitlements": "build/entitlements.mas.plist",
"entitlementsInherit": "build/entitlements.mas.inherit.plist",
"entitlementsLoginHelper": "build/entitlements.mas.loginhelper.plist",
"publish": null
},
which I believe is what is needed to specify that these inner binaries are signed. Perhaps not. Copying others who also had to abandon the docs is not necessarily a guarantee of success. But these things did get me this far.
I've scoured SO and other Internet sites, and of course the poorly written and often obscure official documentation sources, but I can't find the answer to this.
In my travels, I happened across this blog which makes me think that even if I get past this part, there may still be a long bumpy road ahead, and it may turn out to be a dead end.
I notice a lot of developers have commented they chose to abandon the Mac store and instead simply self-publish with their own URL link for download. That seems to be a shame, at the very least, and certainly not my preference.
It makes we wonder if this is even possible. Many of the posts (not to mention the docs) contain outdated information, and few posts are more recent than more than a year. Many things have changed. Maybe everyone threw in the towel. Am I on a futile quest?
I believe I found the answer to the main question of "is it possible?" I may still run into other issues, but I've discovered the main problem.
My Electron project is in a different construction than that of the "Electron-quick-start" template, so the Electron-builder packaging and signing steps do not find what they are looking for in all the normal places.
I am not 100% sure I can successfully reconstruct my project into the 'canonical' format, but that appears to be the solution.
I was able to create a basic quick-start app and upload it to the App Store in a matter of minutes, so all the parts do in fact work. The project just need to be of the right construction.
I'm about to build a chat bot using Microsoft's Bot Framework. I've been transcribing the docs for days to immerse myself in the newness.
It feels like Microsoft wants me to use Composer, but surely Code is King...
I sense that SDK v4 has feature parity with Composer, right?
I can see questions on here about migrating one way or the other. Is this a one way door, in Jeff Bezos speak?
What pitfalls can I expect if I go the coded route? Is this really one of those things that's best done with a GUI?
Thanks in advance.
Of course there's no right answer here. My personal opinion is that Composer makes many "core" features so much easier to create, from dialogs to prompts to using bots as skills. However, as soon as you step out from what Composer considers "standard" (e.g. a dialog loop where you want to maintain some prompt values but reprompt others), you either have to compromise on the features or go to coded. Microsoft is definitely pushing Composer, but I have features that, although maybe not completely necessary, cannot easily (or maybe not at all) be implemented with Composer. I have some of my team exploring Composer bots right now, but for new projects we are still using fully coded bots.
The readme.md at https://github.com/NativeScript/windows-runtime says that the Windows runtime for Nativescript is in proof of concept stage, and then lists what I understand to be very deep language features that are not implemented yet.
The tone on the https://www.nativescript.org/blog/nativescript-runtime-preview-for-windows-10 announcement seems a bit more enthusiastic about the current feature set.
Being able to use Nativescript on Windows Phone (and any other platform) is incredibly appealing.
TJ, a core team member, recently posted on the forums about this:
Hey #NezzaGrey,
Thanks for reaching out, and awesome that you’re liking NativeScript :smile:. >Straight to the point though—we’re not actively working on UWP support because >1) it’s a ton of work to add a new platform and commit to supporting that >platform indefinitely, and 2) we’re not seeing nearly enough demand from our >community to justify taking on that work.
That doesn’t mean that UWP support in NativeScript will never happen, but it’s >not coming in the short term because we’re just not seeing the demand. That can >always change though. I’d encourage you to add your use case to the GitHub >issue open for adding UWP support in NativeScript: >https://github.com/NativeScript/NativeScript/issues/254. Yes, the issue is >somewhat ancient, but we really do pay attention to well-thought-out comments >during roadmap discussions.
I’ll note two other things. First, our initial work on making a Windows runtime >is completely open source and available on GitHub: >https://github.com/NativeScript/windows-runtime. We’d love to have community >?>help to make the new runtime a reality.
Second, one option you have is to build your iOS and Android apps with >NativeScript and Angular, and to use our code sharing approaches (see ?>https://www.nativescript.org/blog/code-sharing-between-web-and-mobile-with->angular-and-nativescript1) to share your Angular code with other apps. You >could take that approach to share Angular code between your NativeScript apps >and your UWP apps if you use something like Electron. This approach isn’t >ideal, as you’d probably prefer to build a completely native UWP app, but it’s >something to consider if you’re open to using Electron.
Anyways, hopefully you found some of this helpful. If you have any other >questions feel free to follow up.
Source: https://discourse.nativescript.org/t/windows-uwp-support/2659/3
We have an existing cross-platform Mobile Application, that also has an accompanying web application, that uses AWS as a back end (RDS for Database, Four server instances, and a Load Balancer to distribute traffic). APK and IPA files are packaged and sent to the stores, while components are retrieved from our server per request, which enables us to reduce the number of store builds, and make the process easier for getting changes out to clients.
We are at the stage of Development that we need to move to another platform to better facilitate our fast growing client base.
Due to the conditions upon our clients, these things must be considered.
must be cross platform (Android / iOS).
must be offline based (users need to be able to access without an internet connection)
must be able to sync with an existing database when there is a connection
requires Authentication
Cloud based (? may not be the right term, but meaning the ability for us to store components on a server and have a device check for updates and download a local copy - enabling us to work and distribute fixes faster)
Ideally compatible with AWS
We are currently looking at Xamarin to facilitate us for this move, however there is a lot of documentation and plugins out there to do all sorts of different things.
As developers, we all have some .net / c# experience, however none particularly with Xamarin.
We have a particular timeline that we need to adhere to (and need to ramp up in the next couple of weeks), and right now are essentially stuck in Limbo in a research phase, as we can't afford to get some things right. We are unable to move forward until we know Xamarin (or some other framework) is able to do everything that we need it to do.
It's hard to escape from the 'code it all ourselves' mentality - especially as we don't know where to start looking in terms of third party packages for Xamarin, and their documentation itself.
A lot of documentation mentions various systems of Azure - we would not be against moving to Azure over AWS if it _had to be done*, but because of the existing infrastructure with AWS - staying with AWS is obviously the preferred option (users need to access the current system whilst we build the new system)
This stack overflow post is to ask for recommendations, comments, or other observations on anything and everything involved with the process in regards to choosing frameworks, design patterns, methodologies, third party packages, etc.
Obviously we would like to use best practises for everything moving forward for optimum scalability and cleaner, more robust code. It's just hard to know where to start!
Any input will be highly appreciated.
Cheers!
edit: I am aware that this is 'asking for recommendations' which is 'specifically off-topic'. I have read the posting guidelines about open ended discussion, and am quite confident that this case is different. There is an underlying problem here, in that we are seeking advice on combinations of frameworks and plugins that are fully compatible with all of the items in the list (above)
Let's try to breakdown each of your requirement and constraint mentioned in the question.
1. You need a offline first architecture (With Sync mechanism)
Xamarin + Azure would make a good comdo for you, as it would support offline storage out of box (With a simple PullAsync call).
Albeit there are AWS SDK available for Xamarin. Here it goes http://docs.aws.amazon.com/mobile/sdkforxamarin/developerguide/setup.html
So the call is yours. There are few other ways to achieve caching offline storage in Xamarin, you can build upon those strategy in your logic. Alternatively there is a very interesting C# library Akavache for caching and offline storage.
2. OAuth 2.0 Authentication
Xamarin has a very good library named Xamarin.Auth. Though I would not say it is very easily extendible at this but there has been some serious work going on from Xamarin on this library.
But I would say it is very easy to use. The apps I have developed so far that includes - Google OAuth, Facebook integration, Microsoft account integration. It worked fine for all of them.
3. Cloud based distribution
There are nice tools available with Microsoft. Which makes the distribution easy. Have a look at https://mobile.azure.com . Also hockeyApp is good for distribution I found.
Where to start:
A very good starting point for you would be https://channel9.msdn.com . Just go there search with keyword "Xamarin" and view some videos. May be all these cross platform dynamics will be much easier for you.
Overall I found Xamarin a cool product to work with. Because anything that can be done in native Objective C/Swift or Java can very well be done in Xamarin using C#.
My team have been battling to make react native play nicely with socket.io for 2 days now, without success. Is anyone currently using the most recent version of react native successfully with socket.io, or is the technology just not ready for commercial environments?
According to the answers to this question, there at least used to be a way to do use socket.io with RN. If that's indeed the case, I'd suggest opening an issue in the github repository, it might be a rather easy fix. I'd also making sure there are no existing issues related to what you're asking about before submitting an issue.
Good luck!