SNOW-723750: Panic with readonly file system

[sfc-gh-pbulawa]

Description

Fixed driver panicking when the temp filesystem is readonly.

Checklist

• Code compiles correctly
• Run make fmt to fix inconsistent formats
• Run make lint to get lint errors and fix all of them
• Created tests which fail without the change (if possible)
• All tests passing
• Extended the README / documentation, if necessary

[mihaitodor]

Thank you for starting this work! Will it be possible to disable the temp file creation mechanism completely for certain workflows? It's not fully clear to me when it's beneficial, especially when the data is streamed from/to a buffer or a file. Also, when one wishes to run a quick INSERT statement with little data, it would be nice to not have to write it to disk first.

@sfc-gh-dszmolka mentioned in this comment that encryption/decription also creates temp files. Will this PR also address that workflow?

Also, it might be worth considering a Snowflake-specific name for the env var which controls the location of this temporary directory, maybe something like SNOWFLAKE_TMPDIR. People might want to have separate directories for various frameworks and relying solely on TMPDIR makes it difficult to distinguish what came from where.

[sfc-gh-pbulawa]

Thank you for starting this work! Will it be possible to disable the temp file creation mechanism completely for certain workflows? It's not fully clear to me when it's beneficial, especially when the data is streamed from/to a buffer or a file. Also, when one wishes to run a quick INSERT statement with little data, it would be nice to not have to write it to disk first.

Removing the usage of the temp folder is not planned.

@sfc-gh-dszmolka mentioned in #700 (comment) that encryption/decription also creates temp files. Will this PR also address that workflow?

The encryption/decryption workflow was not affected with this bug - the creation of the temp file itself does not panic but returns the error which was not properly handled. The driver wanted to assign the error to the nil result which panicked.

Also, it might be worth considering a Snowflake-specific name for the env var which controls the location of this temporary directory, maybe something like SNOWFLAKE_TMPDIR. People might want to have separate directories for various frameworks and relying solely on TMPDIR makes it difficult to distinguish what came from where.

That is a very good idea - we have prepared a feature request to expand the driver with such functionality.

[sfc-gh-igarish]

LGTM