This question already has answers here:
AngularJS 1.4 --> Angular 9 migration vs bigbang rewrite [duplicate]
(2 answers)
Closed 2 years ago.
I have done a ton of research and haven't found anything that has helped me decide what will be the best route (vertical slicing, horizontal slicing, or just a complete rewrite). I am working on a very LARGE program that is very ugly with no comments and need to migrate it over to Angular 8 if possible or at least up to Angular 7. A lot seem to recommend https://angular.io/guide/upgrade however, they don't help too much in migrating to 1.5 first. Does anybody have experience with a large scale migration? Currently, the program is not being used so downtime is no issue.
It doesn't seem like it at first, but a re-write is usually a more cost effective solution than an upgrade. It seems like upgrade would be the fastest time to re-deployment but in my experience if you did the two side-by-side you might find the times similar, except that the migration deployment will have to be all or nothing, where as a re-write means you can deploy with reduced functionality and build up the feature.
More importantly, the on-going maintenance of an upgraded site becomes exponentially harder/time consuming. You're really applying band-aides over the top of previous patches and fixes.
There are new concepts, better native support for directives and controls that we used to use 3rd parties for or roll own own, and it's an entirely new language to understand. Take this opportunity to wipe the slate clean of your solution's technical debt.
Rewrite - Hybrid
Do you need to deploy eveything in one go? Are you interested in Re-Branding?
One thing the Microsoft have done well in the past is the hybrid roll-out of their Preview portals.
The best case study IMO is the Azure Portal.
A few years ago we had a pretty fully featured portal interface for managing Azure assets. This would later become known as the Classic Portal when they started work on an entirely new user experience.
At first release the menu system was largely complete, we could navigate most assets in the new portal but when you came to features that had not been redesigned yet the link took you back to the Classic Portal.
So you could do the same, have the two user interfaces deployed to different URLs, start by making sure the authentication and navigation is largely complete but make all links take the user back to the original interface. Then feature by feature, implement the new interface, but because you can't control everything, keep a button or link on each page that takes the user back to the original implementation until your regression testing confirms that you have reached feature parity.
That is another key take-away from the MS Hybrid approach, significant change like this WILL annoy your users. So while you are in transition, allow the user to choose when they themselves migrate over. Initially MS achieved this at login, the user could login at either of the main urls, and based on your profile you would be redirected to the portal of your choice.
The last step is to restrict access to features in the old interface, by making the navigation and links in the old portal navigate directly into the new interface.
- or less intrusive, add an 'end of life' banner in each page that you have compelted the re-write on in the old site.
Do not confuse this with the Preview Mode in Office 365, the Azure Preview Portal was a ground-up re-write and is still in progress. There are many licensing operations that I still use the Classic Portal for as I still manage some Classic Only azure assets that have not yet been re-deployed.
I would consider the following issues when choosing between Migration/Upgrade and a straight up re-write:
Migrate to AngularJS 1.5 first
This operation is only marginally easier than the upgrade to Angular2+. All of the arguments below apply equally to this process as they do the next upgrade.
One of the reasons to goto 1.5 first is to escape the legacy dependencies that do not have a simple direct upgrade to 2+.
During the upgrade to 1.5 you should consider implementing Component based architecture (if the current code does not already do so)
A key element of components is less configuration and simplifed design, so read this as less to upgrade, less that can go wrong
Components are of course more closely aligned with current Angular implementations, if you do not yet use AngularJS Components, that might be a good iterim step to understand before learning Angular 2+
No Comments
This is a bigger red flag than you think. If the code base is not documented then any sort of maintenance becomes incrementally harder as each time the code must be re-read and re-interpreted before you can affect change.
So if a migration is on the cards, where talking about every line of code at some point needed to be re-read and understood to make sure that it works correctly in the
AngularJS 1+ => Angular2+
While some of the core frameworks and 3rd party libraries can be migrated, most controller javascript files cannot be simply migrated to typescript without a fair amount of effort. It's pretty common in javascript to cut a lot of corners in terms of type definitions and locations of definitions that mean after migration you will spend a lot of time going back through most javascript files one method at a time.
Very large
This is a strong candidate for automation or migration but ultimately it means that the total surface area to test, debug and re-design is also very-large. If the initial migration does not compile, it could be a long road of tweaking before you can get the user interface up so you can start interface testing.
Related
I have a complex appengine service that was written in PHP, now I want to migrate it to Python part by part.
Let's say that my service has 2 parts: /signIn/.... and /data/.... I just want to migrate /signIn/ part first, then /data/ later.
However, since my service is big, so I want to build new /signIn/ part in Python, then use Traffic Splitting to make some A/B Testing on this part.
My problem is that Traffic Splitting can be applied on versions only, so my old and new versions have to be in same module, and same module means that they have to written in same language (I was wrong here, see updated part). But I am migrating from PHP to Python.
What is the best solution for me?
Thanks,
Solution
With Dan Cornilescu's helping, this is what I do:
Split the app into 2 modules: default and old-version.
Dispatch /signIn/ into default module, the rest to old-version module.
Make another version of /signIn/ (default module) in Python
Configure Traffic Splitting to slowly increase requests percent into Python part. This will allow us to test and make sure there is no serious bug happen.
Note:
The /signIn/ part must be default module, since GAE's traffic splitting works at default module only.
I confirmed that we can make 2 versions in different language for a module.
One possible approach is to split your PHP app in modules in a 1st step. It's not a completely wasted effort, most of that will be needed anyways to just allow your app to work in multiple modules, not related to the language change. I suspect this is actually why you can't use A/B testing - mismatch between the modules. Unavoidable.
Once the split in modules is done then you can go on with your 2nd step - switching the language for selected module(s), with A/B testing as you intended.
A more brave approach is to mix the 2 and write the /signin/ module directly in python. On the PHP side you'd just remove the /signin/ portion (part of the earlier mentioned 1st step). Should work pretty well as long as you're careful to only use app language independent means for inter-module communication/operation: request paths, cookies, datastore/memcache keys, etc. A good module split would almost certainly ensure that.
You have testing options other than A/B, like this one: https://stackoverflow.com/a/33760403/4495081.
You can also have the new code/module able to serve the same requests as the old one, side-by-side/simultaneously and using a dispatch.yaml file to finely control which module actually serves which requests. This may allow a very focused migration, potentially offering higher testing confidence.
I'm also not entirely sure you can't actually have 2 versions of the same module in different languages - the versions should be pretty standalone instances, each serving their own requests in their own way using, at the lowest layer, the language-independent GAE infra services. AFAIK nothing stops a complete app re-write and deployment, with the same version or not - I've done that when learning GAE. But I didn't switch languages, it's true. I'd give it a try, but I don't have time to learn a new language right now :)
I have been looking at the various Meanstack frameworks out on the net - and whilst impressed with what they achieve I have one serious concern - the number of files used in a typical stack - meanstack.js uses over 15000 files whilst the bmean example has a modest 1900 in comparison.
The question I am asking myself is would I be happy to put my trust is such a system from a production view point - what happens when something goes wrong how easy is it going to be to find the answer? You can almost bet that when your most important customer logs on it is going to go haywire. Also what happens when Angular version 2 comes along it could require a complete rewrite but by then the stack your using has been customised and difficult to change?
Am I getting over concerned about the technology - my intended approach is to strip the client side code out of the bmean example and rewrite it with my own - at least that way I know (and control) what goes on in the client. Do you think this is the correct way to proceed?
With most systems there is a bit of preparation required before going to production. The same is true with mean.io (using multiple cpu's, improved aggregation, caching, etc etc)
The large number of files is essentially a product of the way npm handles dependencies. Each module is able to define independent versions of the same dependencies thus creating a bit of bloat but at the same time allowing a lot of flexability in nodejs code.
We currently have a number of mean.io projects in production phase and have been very happy with performance and the overall experience.
New releases of the project are scheduled every couple of months, upgrading should not be too much of a problem if you use the package system correctly.
Issues with the project are handled and managed through github issues additional support can be found on our irc (freenode #mean_io) channel as well as on facebook.
For commercial support have a look at the support page
I was starting to write a multi-Map JS library, but I see that Mapstraction does that exactly..
I really would like to use Maptraction but it looks a little old (The commit's on GITHUB) (Not a issue if it is still "supported"), also The tutorial page, the Maps do not show up in my browsers.
Any input is highly appreciated.
Thanks
Kim
Depends what your use case is - if you don't want to be tied in to any one map provider and only need standard features then, yes.
Version 2.1 of Mapstraction is currently being prepared (on the release-2.1 branch) which brings some improvements in behavioural consistency across the providers.
Because of the nature of the library development isn't continuous and tends to be reactive to changes in the underlying providers or issues being raised. That said the community is fairly active and you can sign up to the mailing list via the site.
I am currently investigating possible options of a migration framework/tool. I like the idea of ruby migrations on which the above frameworks are based.
So I am asking for your experience, opinions and maybe a comparison between them. Are you using them in production?
thanks for responses. The goal of this question was to get a feeling about which tools is used most in the developer community but it seems that migrations are not a hot topic here.
Anyway, I have decided to go with MigSharp as the codebase seem to be pretty clean and it is quite easy to handle and had build in support for MS SQL CE. Second runner up would have been FluentMigrator where I was not able to produce a working example for compact edition.
Cheers
I use FluentMigrator in production, and am a longtime contributor to FM. I think your question is to general; be more specific. Also, FM has a google group which is fairly active if you want FM information.
FM is derived from migrator.net, as I recall. It uses a fluent-syntax, and supports multiple databases. We have taken some inspiration from rails migrations, but it's definitely not a port. Worth checking out.
One thing I've learned is not to put your migrations in the same assembly as you app code. Separate them into a migration assembly, and use that for migrating your databases.
Also, you should always work on multiple environments to avoid problems with migrations run straight against production. I always have at least a development and production environment, and most of the time there is a testing environment as well.
I use mig#.
It works well, but you will need to have some guidelines for usage - as migrations can get complicated.
We use sequence number on the end of our migrations rather than a date-time stamp. This is because we don't know when the date time stamp was set (when they begun the source code change-set; just before committing; some time inbetween) different developers could use different approaches.
Names such as Migration_0000034.cs give you plenty of space.
At this point, I would stick with migrator.net. I like the promise of FluentMigrator, but it seems to not have any better active development than migrator.net (see the issues and pull requests that have languished on their github site).
There is also no easy way to do an ExecuteScalar(). I'd add it, but I don't want to create my own fork, and I see no reason that a pull request would actually land in the master. (Execute.WithConnection is an Action so it will fire on demand rather than when I need it to fire)
So for me, I'm heading back to migrator.net.
I'm nearling completion of my first CakePHP-driven website and just saw they're already working on CakePHP 2.0 (not the stable release yet).
My questions:
Is it incredibly time consuming to move to a new version of CakePHP (when it becomes the "stable" release that is)? I know they have migration guides, but - I've never used a framework before, so I've never had to migrate anything.
Do you migrate your code for existing projects, or leave it as is and use the new stable version for future projects only?
Where can I find what version of CakePHP I currently have installed? I've looked at the LICENCE and VERSION files, but cannot find the installed/current version listed in them.
These seem like simple questions, but I greatly appreciate any thoughts/advice - searching this on Google just brings up how-to-migrate pages, not pros/cons...etc.
I've migrated a few sites from CakePHP 1.2 to 1.3. In my experience, it takes 2-3 hours on sites that have 5-10 controllers and no custom plugins, etc. I find I typically only have to change the syntax on a handful of function calls, and when I figure out which ones, it is just a matter of doing a find / replace across the site. Of course it could be more of an issue going from 1.3 to 2.0, but I don't get the sense that it will be an especially drastic API change.
UPDATE: I'm now in the process of migrating to CakePHP 2.0 beta, and thought I should update this, as I'm finding the updates are more extensive and far-reaching than I had assumed when I wrote this. Migration guide here: https://github.com/cakephp/docs/blob/master/en/appendices/2-0-migration-guide.rst
ANOTHER UPDATE: Since people seem to be finding this useful, I just thought I'd point out that Cake now helpfully provides an upgrade shell that does some of the work for you. Note that although the documentation says it will do "most" of the work, I have found there are still quite a few function calls, etc. that will need to be updated manually (see migration guide).
http://book2.cakephp.org/en/console-and-shells/upgrade-shell.html
As dhofstet said, it will all depend on the size and complexity of your site.
Whether you upgrade at all is usually a judgment call, but sometimes you have to (e.g. Cake 1.2 has some code that will break if your host upgrades to PHP 5.3). You certainly wouldn't have the kind of security issues that an old WordPress, Drupal, etc install would have. I have seen some noticeable speed increases with Cake upgrades, so depending on the situation it could be worth the trouble just for that (Cake 2.0 finally drops PHP 4 suppport). Look at the release notes and see if there are things that appeal to you in the new version.
To see your version, in the cake/VERSION.txt file, look at the very last line. It's easy to miss, but it should just be a number, e.g. 1.3.8.
This question is difficult to answer as it depends on the size and complexity of your project(s). The "big" releases (from 1.1 -> 1.2, 1.2 -> 1.3, 1.3 -> 2.0) usually break stuff and so you have to budget some migration work. The migration between "smaller" releases (for example from 1.3.9 to 1.3.10), on the other hand, is usually easy, often it just means to replace the cake folder. In both cases it is useful to have tests.
I migrate the projects which are actively maintained.
You can find the CakePHP version in cake/config/config.php
I'm migrating an app 1.3»2.0rc1 right now and I got no big trouble.
I had to change names of folder/files, eg. app_controller.php » Controller/AppController.php
Follow the migration link (tmp link) http://book2.cakephp.org/en/appendices/2-0-migration-guide.html
plugins/components/.. from various source won't work (at minus, for point 1)
To update the code (which in my case wasn't needed as the app worked well) I've shell-baked a dummy table and looked at differences in code.. It's a good starting point
Authentication/Authorization changed in some config, but requires few changes.
Trees still working
Acl don't. But I'm quite sure it's my fault.
For now it's all, good work!