Zeoslib Portal

aehimself wrote: ↑30.01.2021, 21:06

marsupilami wrote: ↑30.01.2021, 16:04Regarding databases at home: Would it help if there was a computer that you could RDP into to do some things?

RDP - unfortunately no. Plus, I wouldn't accept even a guest account on someone else's server as I wouldn't give any out... sorry, security is my paranoia
I'd need access to databases directly to be able to run tests against different protocols.

Don't worry security wise. It would have been a Windows 10 machine with a Delphi installed - say 10.3 and some of the databases that we use (maybe a selection of SQL Server, MariaDB, PostgreSQL, Firebird, SQLite, Adaptive Server Enterprise, Oracle XE). I did the same for the Jenkins Machine. This Machine would only have had access to the internet and not to our company network. Usually I would prefer something more safe than passwords but I don't know a way to secure RDP sessions with OTPs...

aehimself wrote: ↑30.01.2021, 21:06 Being a Microsoft partner at work I can create an Azure machine for free of charge (they give ~$150 to "play"). I'm thinking on using this to install some RDBMS on it and protect it with a VPN. If I can get it to work I am more than happy to share that with you. That way we could have a couple more instances to test.
What do you guys think, would it benefit the development of Zeos?

I think it really depends on the use cases. What kind of use do you have in mind? Which RDBMS do you have in mind?

Best regards,

Jan

marsupilami wrote: ↑01.02.2021, 11:09I think it really depends on the use cases. What kind of use do you have in mind? Which RDBMS do you have in mind? :)

I have absolutely no details, it just popped up in my mind as a possibility. The machine would be mine (ours... insert communist Bugs Bunny meme here), we could install anything on it we need.
It just needs to be legal :)

aehimself wrote: ↑30.01.2021, 21:12 @ Michael,
I can confirm that your fix works, the test and the original query runs fine now. With 2 parameters I can see 3 binds - which was my original idea.
There are two thing what bugs me:
- Checking the code change I still have no damn clue what you did. Well, you know the code and I don't... anyway, it bugs me.
- The small note I found on OCIBindByPos. It should work as Zeos did it the first place.

Your patch did show me you deeply did dig into the code. Even if it made trouble i appriciate each help. Yet i have not enougth time for the forum. According my change (Quck answer): What oracle describes for duplicate params + OCIBindByPos is somewhat illlogical for me. I can't see the advantages of binding same param more than once as described. Honestly the old code simply missed the correct offsets as you pointed out. So all i did is take care !all! params are filled even if duplicate. OTH we could fix the offsets or use OCIBindByName instead. But this can be done later. Glad to hear it working now.

@Michael,

Please see this topic. I think the issue is created because of the change of the parameter interpretation...

Edit: Please ignore. As it turns out, it was a bug in Zeos, see here.