Convert RZ to TAR.GZ Online (Trading a 900 MB Window for a 32 KB One)

What Happens When Rzip's Layer Comes Off and Gzip's Goes On

Rzip, written by Andrew Tridgell (also the creator of Samba and rsync) with its last stable release dating to February 2006, compresses in two stages: a long-range matching pass across a window reaching up to 900 MB, then bzip2 on whatever remains. Gzip, based on the DEFLATE algorithm specified in RFC 1951, uses a sliding window of only 32 KB — several orders of magnitude smaller. Because rzip typically compresses an existing tar archive to begin with, converting RZ to TAR.GZ means decompressing rzip's layer to recover that tar archive, then compressing it fresh with gzip instead.

That window-size gap is the single biggest factor in how the converted file's size compares to the original. Gzip simply cannot see repetition beyond 32 KB apart, no matter how obviously repeated the data is at a larger scale — a limitation rzip's matching pass was built specifically to work around.

This conversion tends to happen for compatibility reasons rather than compression-quality reasons: a .tar.gz file is one of the most widely recognized archive extensions in open-source software distribution, while .rz remains obscure enough that a recipient without rzip specifically installed has no straightforward way to open it at all.

Why TAR.GZ Remains the Format Most Systems Expect by Default

Gzip's format is fully specified in RFC 1952, an openly published standard that any developer can implement from scratch without licensing concerns, and GNU tar's -z flag has offered built-in gzip support for decades, making tar czf output.tar.gz directory a single, familiar command on virtually any Linux or macOS system. Rzip, by contrast, never achieved that same default-tool status — its own project page frames it as a tool for people who specifically "have ram to burn and want the best possible compression rate," not a general-purpose everyday compressor.

This difference in adoption is a real, practical reason conversions in this direction happen at all: a .tar.gz file will open on essentially any system without a second thought, while a .rz or .tar.rz file requires the recipient to specifically have rzip or PeaZip installed — a real barrier when sharing a file outside a controlled environment.

Gzip's own format also carries a documented limitation of its own: its trailer field stores the uncompressed size modulo 4 GB, meaning files above that threshold can report an incorrect size after decompression on some older tools, though the actual data still decompresses correctly on modern gzip implementations that read the stream length directly rather than trusting the trailer alone.

Gzip's DEFLATE also compresses noticeably faster than bzip2's block-sorting approach on most hardware, since Huffman coding combined with a small sliding window is computationally simpler than the Burrows-Wheeler transform bzip2 applies to each of its roughly 900 KB blocks — a real, practical speed advantage for anyone converting a very large file where compression time matters more than squeezing out the last few percent of size.

What Converting RZ to TAR.GZ Gains and What It Costs

• Gain — universal, no-install compatibility: gzip ships by default on essentially every Linux distribution and macOS, and Windows tools like 7-Zip open .tar.gz without configuration; .rz needs rzip or a specific compatible tool.
• Gain — faster decompression on the receiving end: DEFLATE's smaller window and simpler algorithm decompress faster than bzip2's block-sorting approach, which rzip relies on for its second stage.
• Lose — rzip's long-range matching advantage entirely: on the project's own Samba source tree benchmark, rzip plus bzip2 reached 9.55x compression versus bzip2 alone's 4.78x; gzip's DEFLATE, with its much smaller window, typically falls further behind still on data with that kind of far-apart redundancy.
• Lose — the original compression work: this conversion means fully decompressing rzip's output and recompressing from scratch, not building incrementally on what rzip had already found.
• Keep — Unix metadata inside the tar layer: permissions, ownership, and modification times recorded in tar's own header survive the round trip unchanged, since neither rzip's nor gzip's compression touches that structure.
• Gain — quicker decompression for the eventual recipient: a smaller matching window and simpler entropy coding mean a .tar.gz file typically extracts faster than a .tar.bz2 or a rebuilt .tar.rz would on the same hardware.

Which Tools Handle Decompression and Recompression Here

The rzip binary itself, GPL-licensed and packaged in Debian and Ubuntu's repositories, decompresses the source with rzip -d filename.tar.rz, recovering the plain tar archive. From there, GNU tar's -z flag or the standalone gzip command compresses that recovered archive into the final .tar.gz — both are standard, preinstalled tools requiring no additional setup on Linux or macOS.

On Windows, 7-Zip and PeaZip both build .tar.gz archives directly through their graphical interfaces, and PeaZip specifically also lists native support for reading the original .rz format — making it the more practical single-application choice for handling both halves of this conversion without needing a separate command-line rzip build.

There's no dedicated Windows binary maintained by the rzip project itself, so decompressing the .rz source on Windows depends specifically on PeaZip rather than the original author's own software — a genuine asymmetry compared with gzip, which has long had multiple independent, actively maintained Windows ports.

Real Problems Reported Converting Rzip Output to TAR.GZ

A common complaint involves someone converting a large backup file from RZ to TAR.GZ expecting a comparable file size, then finding the TAR.GZ noticeably larger — the consistent explanation is that the original data had significant repetition spread across large distances, exactly the pattern rzip's 900 MB window catches and gzip's 32 KB window structurally cannot.

A second documented issue involves a recipient unable to open a shared .rz file at all, since few mainstream tools besides PeaZip support it — a direct consequence of rzip staying a specialist Linux/Unix utility after development effectively stopped in 2006, in contrast with gzip's near-total ubiquity across every major operating system.

A third reported pattern involves confusing the original rzip with its actively maintained but incompatible successor, lrzip — a file compressed with one cannot be decompressed with the other despite the shared name and similar long-range-matching concept, producing an error rather than a usable tar archive to recompress into TAR.GZ.

A fourth pattern shows up around build scripts and package managers that specifically check for a .tar.gz extension before proceeding — pointing the same script at a .rz or .tar.rz file instead fails that check outright, a maintenance issue tied to the script's narrow assumption about file naming rather than anything wrong with the underlying data once it's actually decompressed.

RZ and TAR.GZ Compared on the Details That Matter

Questions About Converting RZ Files to TAR.GZ

Why is my converted TAR.GZ file bigger than the original RZ file?
Gzip's DEFLATE algorithm uses a 32 KB matching window, far smaller than rzip's 900 MB — if the original data had repetition spread across large distances, gzip simply cannot find or exploit it the way rzip's matching pass did.

Is there a single command to convert RZ straight to TAR.GZ?
Not a dedicated one. The standard route decompresses with rzip first, then recompresses the recovered tar archive using GNU tar's -z flag or the standalone gzip command.

Does TAR.GZ open on more systems than RZ does?
Yes, by a wide margin. Gzip ships by default on virtually every Linux distribution and macOS, while opening an .rz file typically requires rzip itself or PeaZip specifically, neither of which comes preinstalled. Windows also has multiple independently maintained gzip ports going back many years, unlike rzip, which has no dedicated Windows build from its own developers.

Will Unix file permissions survive this conversion?
Yes. Permissions, ownership, and modification times are recorded in tar's own header, and neither rzip's nor gzip's compression touches that structure, so they pass through the round trip unchanged, including symlink targets and sparse-file markers where GNU tar recorded them originally.

Does gzip's file-size limit affect files converted from RZ?
Potentially, for files over 4 GB. Gzip's trailer stores the uncompressed size modulo 4 GB, which can report an incorrect size on some older tools, though modern implementations generally decompress the data correctly regardless.

Is gzip faster than rzip's bzip2-based approach?
Usually yes, for the compression step itself. DEFLATE's smaller window and simpler Huffman coding process data faster than bzip2's block-sorting transform, though rzip's own first-stage matching can still make its overall process competitive on highly redundant files.