SQLite 3.35

[ericsink]

Mentioned in dotnet/efcore#23552

[springy76]

There is already

• 2021-01-20 (3.34.1)
• 2021-03-12 (3.35.0)
• 2021-03-15 (3.35.1)
• 2021-03-17 (3.35.2)
• 2021-03-26 (3.35.3)
• 2021-04-02 (3.35.4)
• 2021-04-19 (3.35.5)

Why there is no e_sqlite for each and every release of sqlite (within a few days)? Is there much work to do for each release, or what am I missing?

[ericsink]

Yeah I'm a bit behind on these, aren't I?

In part, I've been waiting for 3.35 to settle down.

[DannyBoyk]

We just got hit by this as well. We have to handle databases that we don't really have any control over, and someone decided to use the new 3.31 Generated Columns feature. Do you anticipate a release anytime soon? Not sure what other options we have to get the latest SQLite version. We've build everything around SQLitePCL.raw.

[springy76]

@DannyBoyk SQLitePCLRaw.bundle_e_sqlite3 2.0.4 comes with 3.33.0 - 2.0.5-pre20210119130047 with 3.34.0

[DannyBoyk]

@DannyBoyk SQLitePCLRaw.bundle_e_sqlite3 2.0.4 comes with 3.33.0 - 2.0.5-pre20210119130047 with 3.34.0

Huh... what about bundle SQLitePCLRaw.bundle_e_sqlcipher? We updated to 2.0.4 and still had a failure opening a database using the GENERATED ALWAYS directive. I assumed the information here about v3.28 was still accurate: https://github.com/ericsink/SQLitePCL.raw/blob/master/v2. DB Browser for SQLite was able to open it running version 3.32.2, so I assumed we were still below 3.31.

Error:

[06:51:54 WRN] Sqlite database appears corrupt, attempting to repair it. Quick check returnCode: 11, errorMessage: SQLite Error 11: 'malformed database schema (events) - near "GENERATED": syntax error'., details:
[06:51:54 WRN] Table: events missing column names.
[06:51:54 ERR] Failed to repair corrupt database. returnCode: 1, errorMessage: SQLite Error 1: 'near "GENERATED": syntax error'.

The log messages specifically reflect our code running an integrity check, detecting a failure, and attempting to repair the database, respectively.

[DannyBoyk]

I see that this is a known issue with sqlcipher. #382