std.tar: add support for file path in pax attributes

[189900]

Handles .extended_header type to parse PAX attributes and check if they override the path of the next file. Increases maximum file path length from 255 to 1014std.fs.MAX_PATH_BYTES.

Fixes#15342

[truemedian]

Is there a reason you chose 1014 specifically instead of something like std.fs.MAX_PATH_BYTES?

[189900]

I went with 1014 to maintain the existing read buffer characteristics and avoid extra overhead.

We can definitely switch to std.fs.MAX_PATH_BYTES if that would be more preferable.

[truemedian]

PAX path attributes technically have no length limitation, so we need to allocate as much space as we can reasonably handle so that the issue of not being able to handle long file paths doesn't appear again. Any path > MATH_PATH_BYTES can't be sent to the os anyway, so it's a good upper limit.

[squeek502]

Going to link to my idea for parsing one-component-at-a-time to support arbitrarily long file paths without needing an allocator: #15382 (comment)

Just using a buffer length of MAX_PATH_BYTES might be good enough without the added complication of the one-component-at-a-time strategy, though.

[189900]

Thank you both. Updated the branch to support std.fs.MAX_PATH_BYTES. A sample tar with a long path: 4k.tar.gz

[189900]

Handles .extended_header type to parse PAX attributes and check if they override the path of the next file. Increases maximum file path length from 255 to 1014std.fs.MAX_PATH_BYTES.

Fixes#15342

[truemedian]

Is there a reason you chose 1014 specifically instead of something like std.fs.MAX_PATH_BYTES?

[189900]

I went with 1014 to maintain the existing read buffer characteristics and avoid extra overhead.

We can definitely switch to std.fs.MAX_PATH_BYTES if that would be more preferable.

[truemedian]

PAX path attributes technically have no length limitation, so we need to allocate as much space as we can reasonably handle so that the issue of not being able to handle long file paths doesn't appear again. Any path > MATH_PATH_BYTES can't be sent to the os anyway, so it's a good upper limit.

[squeek502]

Going to link to my idea for parsing one-component-at-a-time to support arbitrarily long file paths without needing an allocator: #15382 (comment)

Just using a buffer length of MAX_PATH_BYTES might be good enough without the added complication of the one-component-at-a-time strategy, though.

[189900]

Thank you both. Updated the branch to support std.fs.MAX_PATH_BYTES. A sample tar with a long path: 4k.tar.gz

[189900]

Handles .extended_header type to parse PAX attributes and check if they override the path of the next file. Increases maximum file path length from 255 to 1014std.fs.MAX_PATH_BYTES.

Fixes#15342

[truemedian]

Is there a reason you chose 1014 specifically instead of something like std.fs.MAX_PATH_BYTES?

[189900]

I went with 1014 to maintain the existing read buffer characteristics and avoid extra overhead.

We can definitely switch to std.fs.MAX_PATH_BYTES if that would be more preferable.

[truemedian]

PAX path attributes technically have no length limitation, so we need to allocate as much space as we can reasonably handle so that the issue of not being able to handle long file paths doesn't appear again. Any path > MATH_PATH_BYTES can't be sent to the os anyway, so it's a good upper limit.

[squeek502]

Going to link to my idea for parsing one-component-at-a-time to support arbitrarily long file paths without needing an allocator: #15382 (comment)

Just using a buffer length of MAX_PATH_BYTES might be good enough without the added complication of the one-component-at-a-time strategy, though.

[189900]

Thank you both. Updated the branch to support std.fs.MAX_PATH_BYTES. A sample tar with a long path: 4k.tar.gz

[189900]

Handles .extended_header type to parse PAX attributes and check if they override the path of the next file. Increases maximum file path length from 255 to 1014std.fs.MAX_PATH_BYTES.

Fixes#15342

[truemedian]

Is there a reason you chose 1014 specifically instead of something like std.fs.MAX_PATH_BYTES?

[189900]

I went with 1014 to maintain the existing read buffer characteristics and avoid extra overhead.

We can definitely switch to std.fs.MAX_PATH_BYTES if that would be more preferable.

[truemedian]

PAX path attributes technically have no length limitation, so we need to allocate as much space as we can reasonably handle so that the issue of not being able to handle long file paths doesn't appear again. Any path > MATH_PATH_BYTES can't be sent to the os anyway, so it's a good upper limit.

[squeek502]

Going to link to my idea for parsing one-component-at-a-time to support arbitrarily long file paths without needing an allocator: #15382 (comment)

Just using a buffer length of MAX_PATH_BYTES might be good enough without the added complication of the one-component-at-a-time strategy, though.

[189900]

Thank you both. Updated the branch to support std.fs.MAX_PATH_BYTES. A sample tar with a long path: 4k.tar.gz

[189900]

Handles .extended_header type to parse PAX attributes and check if they override the path of the next file. Increases maximum file path length from 255 to 1014std.fs.MAX_PATH_BYTES.

Fixes#15342

[truemedian]

Is there a reason you chose 1014 specifically instead of something like std.fs.MAX_PATH_BYTES?

[189900]

I went with 1014 to maintain the existing read buffer characteristics and avoid extra overhead.

We can definitely switch to std.fs.MAX_PATH_BYTES if that would be more preferable.

[truemedian]

PAX path attributes technically have no length limitation, so we need to allocate as much space as we can reasonably handle so that the issue of not being able to handle long file paths doesn't appear again. Any path > MATH_PATH_BYTES can't be sent to the os anyway, so it's a good upper limit.

[squeek502]

Going to link to my idea for parsing one-component-at-a-time to support arbitrarily long file paths without needing an allocator: #15382 (comment)

Just using a buffer length of MAX_PATH_BYTES might be good enough without the added complication of the one-component-at-a-time strategy, though.

[189900]

Thank you both. Updated the branch to support std.fs.MAX_PATH_BYTES. A sample tar with a long path: 4k.tar.gz

[189900]

Handles .extended_header type to parse PAX attributes and check if they override the path of the next file. Increases maximum file path length from 255 to 1014std.fs.MAX_PATH_BYTES.

Fixes#15342

[truemedian]

Is there a reason you chose 1014 specifically instead of something like std.fs.MAX_PATH_BYTES?

[189900]

I went with 1014 to maintain the existing read buffer characteristics and avoid extra overhead.

We can definitely switch to std.fs.MAX_PATH_BYTES if that would be more preferable.

[truemedian]

PAX path attributes technically have no length limitation, so we need to allocate as much space as we can reasonably handle so that the issue of not being able to handle long file paths doesn't appear again. Any path > MATH_PATH_BYTES can't be sent to the os anyway, so it's a good upper limit.

[squeek502]

Going to link to my idea for parsing one-component-at-a-time to support arbitrarily long file paths without needing an allocator: #15382 (comment)

Just using a buffer length of MAX_PATH_BYTES might be good enough without the added complication of the one-component-at-a-time strategy, though.

[189900]

Thank you both. Updated the branch to support std.fs.MAX_PATH_BYTES. A sample tar with a long path: 4k.tar.gz

[189900]

Handles .extended_header type to parse PAX attributes and check if they override the path of the next file. Increases maximum file path length from 255 to 1014std.fs.MAX_PATH_BYTES.

Fixes#15342

[truemedian]

Is there a reason you chose 1014 specifically instead of something like std.fs.MAX_PATH_BYTES?

[189900]

I went with 1014 to maintain the existing read buffer characteristics and avoid extra overhead.

We can definitely switch to std.fs.MAX_PATH_BYTES if that would be more preferable.

[truemedian]

PAX path attributes technically have no length limitation, so we need to allocate as much space as we can reasonably handle so that the issue of not being able to handle long file paths doesn't appear again. Any path > MATH_PATH_BYTES can't be sent to the os anyway, so it's a good upper limit.

[squeek502]

Going to link to my idea for parsing one-component-at-a-time to support arbitrarily long file paths without needing an allocator: #15382 (comment)

Just using a buffer length of MAX_PATH_BYTES might be good enough without the added complication of the one-component-at-a-time strategy, though.

[189900]

Thank you both. Updated the branch to support std.fs.MAX_PATH_BYTES. A sample tar with a long path: 4k.tar.gz

[189900]

Handles .extended_header type to parse PAX attributes and check if they override the path of the next file. Increases maximum file path length from 255 to 1014std.fs.MAX_PATH_BYTES.

Fixes#15342

[truemedian]

Is there a reason you chose 1014 specifically instead of something like std.fs.MAX_PATH_BYTES?

[189900]

I went with 1014 to maintain the existing read buffer characteristics and avoid extra overhead.

We can definitely switch to std.fs.MAX_PATH_BYTES if that would be more preferable.

[truemedian]

PAX path attributes technically have no length limitation, so we need to allocate as much space as we can reasonably handle so that the issue of not being able to handle long file paths doesn't appear again. Any path > MATH_PATH_BYTES can't be sent to the os anyway, so it's a good upper limit.

[squeek502]

Going to link to my idea for parsing one-component-at-a-time to support arbitrarily long file paths without needing an allocator: #15382 (comment)

Just using a buffer length of MAX_PATH_BYTES might be good enough without the added complication of the one-component-at-a-time strategy, though.

[189900]

Thank you both. Updated the branch to support std.fs.MAX_PATH_BYTES. A sample tar with a long path: 4k.tar.gz