You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Pharo windows fails to connect to a postgresql database hosted in azure. This is a managed instance, so there is no user setup required or even possible for some elements.
Pharo linux and Pharo mac connect without any issues via the P3 postgresql driver, but pharo windows returns an error: -5. After some searching, the error comes from the SqueakSSL.c library from the pharo VM.
The error corresponds to SQSSL_GENERIC_ERROR and this error is used in sqWin32SSL.c at 8 different locations, so it doesn't pinpoint the source of the error.
Using PharoConsole.exe --logLevel=5 (for trace) I was able todetermine that the error is from this part of the code:
I propose (request) 2 changes to the SqueakSSL library:
Increase the number of errors defined in SqueakSSL.h library and point a unique error to each section returning a different error number. This way the user can be better informed. Using the error numbers, the corresponding SSL object in pharo can return a better textual description of the error instead of returning a "-5."
Instead of logging the errors at the trace level, change to error level so that it is more apparent that something is failing in the execution of the SSL connection.
This does not solve the connection problem, but it will help with debugging this and future SSL features.
The text was updated successfully, but these errors were encountered:
sure! and thanks for looking into this issue!
for 1)
SQSSL_GENERIC_ERROR=-5 is defined in SqueakSSL.h and is used several times in sqWin32SSL.c
The idea is to add more error #defines in SqueakSSL.h (i.e -6, -7, -8, ...) and use a different error for the different sections in sqWin32SSL.c. ,Then these numbers can be translated to user readable errors in the #ZdcPluginSSLSession class in pharo.
With the current approach, errors belonging to different situations report the same generic (-5) error.
For 2)
logTrace in sqWin32SSL.c is being used for 2 objectives, tracing and error reporting.
I think in the following lines, there should be a logError : 208, 240, 266, 591, 594, 608, 706, 777, 870, 889, 975, 997, and 1018.
When doing a PR, does the CI create a VM that can be tested?
These changes will allow improving error messages in pharo but does not solve my connection issue; I need to get a VM that can be run under debugging in VS.
Pharo windows fails to connect to a postgresql database hosted in azure. This is a managed instance, so there is no user setup required or even possible for some elements.
Pharo linux and Pharo mac connect without any issues via the P3 postgresql driver, but pharo windows returns an error: -5. After some searching, the error comes from the SqueakSSL.c library from the pharo VM.
The error corresponds to SQSSL_GENERIC_ERROR and this error is used in sqWin32SSL.c at 8 different locations, so it doesn't pinpoint the source of the error.
Using PharoConsole.exe --logLevel=5 (for trace) I was able todetermine that the error is from this part of the code:
I propose (request) 2 changes to the SqueakSSL library:
Increase the number of errors defined in SqueakSSL.h library and point a unique error to each section returning a different error number. This way the user can be better informed. Using the error numbers, the corresponding SSL object in pharo can return a better textual description of the error instead of returning a "-5."
Instead of logging the errors at the trace level, change to error level so that it is more apparent that something is failing in the execution of the SSL connection.
This does not solve the connection problem, but it will help with debugging this and future SSL features.
The text was updated successfully, but these errors were encountered: