Copy ASIO4ALL settings from old version#1260
Conversation
ann0see
left a comment
There was a problem hiding this comment.
Not yet ready; added some explanations
| @@ -212,6 +218,30 @@ Section "Install_64Bit" INST_64 | |||
| MessageBox MB_OKCANCEL|MB_ICONEXCLAMATION "$(OLD_VER_REMOVE_FAILED)" /sd IDCANCEL IDOK continueinstall | |||
| goto quit | |||
| ${EndIf} | |||
There was a problem hiding this comment.
This code will only be executed if we find jamulus in the 32 bit prog. files folder and the current install is 64 bit.
|
Feel free to edit the code. |
|
The latest commit makes it very simple. And seems work. We're not checking for anything, so probably it would also overwrite some other things. |
|
Tests so far with new code:
Still todo:
|
windows/installer.nsi
Outdated
|
|
||
| ; ask to move old ASIO4ALL registry configuration | ||
|
|
||
| MessageBox MB_YESNO|MB_ICONEXCLAMATION "If your're using ASIO4ALL: Do you want to copy ASIO4ALL settings from the previous version if possible? We will do this for every user." /sd IDYES IDNO continueinstall |
There was a problem hiding this comment.
Do we need this message?
There was a problem hiding this comment.
I can't think of any reason why users would not want it.
As a user, I'd expect Jamulus+ASIO4ALL to work the same after the update than before. This logic tries to do that.
The logic is supposed to be non-destructive, i.e. it should not overwrite any configs (we should be really sure about that).
So my tendency would be to not add a (possibly confusing) message. Would also have the benefit of not needing another translators ping...
There was a problem hiding this comment.
Ok. Just did a force push without this change. I think now we'll just need to do some tests.
|
Still one edge case: What should happen if an user now sets a different installation path? Do we need to handle this? |
That would have always been a problem, wouldn't it? I think we could add a message on the page where the target path is chosen and warn about it ("If you change the installation path from previous installations, you may have to re-configure ASIO4ALL."). However, as this is not a regression as far as I understand) and because it would benefit from being translated, I'd suggest to postpone that part for 3.7.1? |
I think so. Not sure, but the installation path couldn't even be saved for an upgrade in the old installer, could it? |
softins
left a comment
There was a problem hiding this comment.
Appears to work ok on my W7 64-bit:
- Installed 3.6.2
- Made sure key 7A49ECC9 did not exist and 8A9E7A56 did exist.
- Installed artifact from this PR. Was warned about removing the 32-bit version, which I accepted.
- Checked registry and confirmed everything in 8A9E7A56 had been copied to 7A49ECC9.
- Only one user on this PC had ever used Jamulus, so couldn't check for others.
|
Logged in as admin:
I think this is the most important check we need to make. |
hoffie
left a comment
There was a problem hiding this comment.
I can't really comment on the NSH and NSI code as I have zero experience with it. I can't test on Windows either, sadly.
However, I'm confident that we should have this change and logic-wise it looks sounds.
I'd be happy if yet another person could test the installer, either explicitly or through another rc.
Co-authored-by: Christian Hoffmann <[email protected]>
|
Before merging this, I'd like to have feedback from @gilgongo everything should work as expected. |
|
I can install it on my machine? It has old settings from a previous install. Where do I download it? |
|
I think the windows compilation process isn't finished yet, so you'll need to wait a bit. Afterwards you can find the binary on the checks page of this PR. |
|
I've merged this now since I think it works as expected. |
Fixes #1244