odbc-driver | Laravel 4 ODBC Driver | Data Migration library

Why did this happen? This is a result of Microsoft lacking packages that support arm64 architecture (Apple silicon is ARM-based). Unbeknownst to me, this is apparently a detail that is observed at a lower level in Docker while requesting packages in the fashion described in the original question.

Solution The solution is to specify a supported platform architecture when building the image. This can be done as an argument to docker build, or it can be added between FROM and your image's identifier in your Dockerfile.

I experienced success using linux/amd64 as shown:

• From the command line

  • docker build --platform=linux/amd64 .

• In the Dockerfile

  • FROM --platform=linux/amd64 public.ecr.aws/docker/library/python:3.9.10-slim-buster

The problem was that the Databricks SQL driver does not yet support ARM, which my laptop and Docker container was building for. See ('01000', "[01000] [unixODBC][Driver Manager]Can't open lib '/opt/simba/spark/lib/64/libsparkodbc_sb64.so' : file not found (0) (SQLDriverConnect)")

I updated my repo at https://github.com/rlorenzo/databricks_php with a working example and forced the container to build an x86 version.

Will update my repo if Databricks ever supports ARM.

Hope this helps someone in the future.

not quite sure about your specific problem here. Especially since you´re not able to delete it manually and a bit missing info, like

• is older version of the same db also run via ODBC
• is the aforementioned error occuring in both failed cases.

But I solved the multiple users access same data problem in most cases by checking my code with regard to the following:

• open recordsets that are not closed before the delete command.

What the best way to install "Microsoft Access Driver (*.mdb, *.accdb)" in this [Alpine Linux] Docker image?

You can't. Microsoft only provides "Microsoft Access Driver (*.mdb, *.accdb)" for the Windows platform (unlike their ODBC driver for SQL Server which is available for Windows, Linux, and Mac).

Some alternatives are discussed here:

Working with an Access database in Python on non-Windows platform (Linux or Mac)

The easiest way is using this repo

It's as easy as

I'll attempt to answer my own question as best I can. Let me first look at implementing the factory pattern via classes.

The two terms subclassing and subtyping are often mentioned within the context of inheritance. While the former implies a syntactic relationship through implementation reuse (implementation inheritance), the latter implies a semantic "is-a" relationship (interface inheritance). The two concepts are often conflated in Python, but when I ask whether MyURL objects are or aren't URL objects, I'm referring to the semantic relationship between the two.

Of course, when I subclass URL in my code example above, I am indeed creating a subtype of URL that satisfies the Liskov Substitution Principle (LSP): I added a few methods (i.e. I specialized URL), but I can still pass MyURL instances to SQLAlchemy's create_engine function and nothing breaks. That's because MyURL implements the complete interface of its (generalized) superclass.

What I am really trying to achieve, though, is for MyURL to not only add those few methods and attributes but to also only possess (or expose) a subset of the methods of its superclass, in an attempt to disable ways to create URL strings that would be incompatible with SQL Server. Others have asked about removing superclass methods in subclasses (see here and here for example), but doing so will violate the LSP as well as the "is-a" relationship between the two classes.

So I suppose that inheritance through subclassing is in fact not what I should be doing here.

Delegation is another example of implementation reuse whereby not a class "blueprint" is shared but an instance of a class. It's therefore more of a "has-a" relationship. Specifically, in my code example I'm doing an implicit delegation, since I'm not passing URL or an instance thereof as a parameter to MyURL's methods. URL.create is a class method, so I can access it directly. In fact, since SQLALchemy's URLs are themselves a subclass of (immutable) tuples I wouldn't even be able to create my specialized version after instantiating them.

Some of my confusion about what point there is to having instances of MyURL stemmed from the fact, that I was still to much focused on a "is-a" relationship. Realizing that this is not the case makes it clearer that what MyURL actually is is a factory class to create URLs. I could rename it to MyURLFactory to make that distinction clearer.

I could even remove the @classmethod decorators. To use MyURL I would then have to instantiate it before use (although I'm not sure what the benefit of that would be):

So, multiple things were not configured in the right way in my Dockerfile, here's the patch :

Thanks to Wernfried Domscheit I finally got it to work. Thanks man! Here is how I did it:

1. Install Visual Studio

2. Download Version 18.5 Basic Package / ODBC Package

   https://www.oracle.com/de/database/technologies/instant-client/microsoft-windows-32-downloads.html

   https://www.oracle.com/de/database/technologies/instant-client/winx64-64-downloads.html#ic_winx64_inst

3. unzip both in the same directory (I choose C:\oracle\odbc\x86(x64)\instantclient_18_5) and install the driver via cmd as admin: C:\oracle\odbc\x86\instantclient_18_5>odbc_install.exe

4. create C:\oracle\admin and put tnsnames.ora in

If your macos is on the x64 architecture, you can use below method to solve issues.

If you use M1, it's not support now.

Solution for x64 architecture.

Run below code, it should useful to you.