Create tar
Bundle files into a plain (uncompressed) .tar, on your device. Hand-rolled USTAR, so see Extract tar for what it does not handle.
Runs on your device. The file is never uploaded.
Create tar bundles the files you drop into one uncompressed USTAR .tar, archive.tar unless outputName says otherwise. Each entry goes in as a regular file with mode 0644, uid and gid both 0, and an epoch timestamp, so the same inputs always produce identical bytes. A path too long for the 100-byte name and 155-byte prefix fields is refused.
Questions
Is the tar compressed?
No. This writes a plain, uncompressed USTAR archive, which is what tar itself does without a compression flag. To get a .tar.gz, run Gzip compress on the result, which appends .gz and gives you archive.tar.gz. The order matters: tar first, then gzip, because gzip compresses a single stream and knows nothing about file names.
What does it write into the header fields?
Fixed values, so the output is deterministic: mode 0644, uid and gid 0, and a modification time of the epoch. Every entry is written as a regular file with typeflag 0. Running the same files through twice gives you byte identical archives, which is useful when you are comparing or checksumming builds.
Are there limits on the file names?
Yes, the USTAR limits. A path is split into a final segment of at most 100 bytes and a prefix of at most 155 bytes, and a path that cannot be split that way is refused with a message saying so. The writer never emits GNU long name or PAX extended headers, which is why archives it makes always read back with Extract tar.
What if two files have the same name?
The run is refused with "two input files are both named" and the name. An archive quietly missing a file you selected is a worse outcome than an error, so it stops and asks you to rename one. The check runs as the archive is being built and before anything is handed back, so you never get a partial tar.
What is the archive called?
archive.tar by default, changed with the outputName option, which is used exactly as typed. The archive ends with the standard two zero blocks that mark the end of a tar, so ordinary tar tools read it without complaint. Entries keep the names of the files you dropped, split across the USTAR name and prefix fields when a path is long.
Are my files uploaded?
No. Every file is read into memory and the archive is assembled there in your browser. Nothing is sent anywhere. Because the inputs and the finished tar are all in memory at once, and a tar adds padding to a 512 byte boundary per entry, the total you can pack is bounded by your device.