BUG 1.4: do not suppress errors when closing file handles

[twoertwein]

xref #47153 (comment), closes #47136

[simonjayhawkins]

@twoertwein can you open against main instead and we will merge this first, then merge main on the other PR

[twoertwein]

@twoertwein can you open against main instead and we will merge this first, then merge main on the other PR

These changes will fail on main because _BytesTarFile was causing errors. I'm not entirely sure why, maybe super().close() will call close on both superclasses which then caused a ValueError.

[simonjayhawkins]

right. so this PR is effectively a manual partial backport of #47153 (opened prematurely) and not broken off from #47153

I suppose that's OK (although not how we normally do things) if there are other issues on main.

the other PR should be merged first then?

[twoertwein]

right. so this PR is effectively a manual partial backport of #47153 (opened prematurely) and not broken off from #47153

#47153 also de-duplicates the write buffer across _BytesTarFile and _BytesZipFile which is technically not needed for this bug.

the other PR should be merged first then?

When seeing it as a backport, it would make sense to merge the other one first.

[twoertwein]

maybe a cleaner option: I do not remove the try-catch in #47153 (just fix the underlying bug+cleanup), open this PR against main after #47153 is merged

[twoertwein]

maybe a cleaner option: I do not remove the try-catch in #47153 (just fix the underlying bug+cleanup), open this PR against main after #47153 is merged

I removed the bug fix in #47153

[simonjayhawkins]

maybe a cleaner option: I do not remove the try-catch in #47153 (just fix the underlying bug+cleanup), open this PR against main after #47153 is merged

sgtm

[twoertwein]

Closing this one as I don't know how to change the target branch. Opened #47165.