- 27 Dec, 2017 9 commits
-
-
Mark Adler authored
-
Mark Adler authored
-
Mark Adler authored
Previously pigz would return an exit code of 0 if there were no serious errors, e.g. I/O errors or insufficient memory. Now pigz will return 1 if it encounters any issues, such as invalid or unrecognized compressed data, file names without the expected suffix, trying to overwrite a file, etc. Since pigz can process multiple input files, it will return an exit code of 0 if there were no issues, and will return an exit code of 1 if one or more of the inputs had an issue.
-
Mark Adler authored
An empty suffix with forced compression would result in both the original and compressed file being deleted.
-
Mark Adler authored
Avoid warning for ignoring chown() return value on some glibc's, and change compiler from cc to gcc. Sometimes the (void) before chown() is not sufficient to signal the programmer's intent to ignore the return value. (Thank you return-value police.) This commit kills that gcc warning just for copymeta(). Using pragma then required ignoring a warning for versions of gcc that don't recognize it (when using -Wall). Finally, gcc is used instead of cc for platforms that don't make that equivalency.
-
Mark Adler authored
-
Mark Adler authored
This adds an option present in gzip to improve interoperability.
-
Mark Adler authored
Change command line processing to go through all of the options and associated parameters first, and then go through the command line again, processing the file names using the resolved options. In addition, if "-" follows a "--", then consider the "-" to be a file name instead of stdin. Both of these changes are to be more compatible with gzip, and are not compatible with the previous versions of pigz, which would process the options and file names in the sequence they appear.
-
Mark Adler authored
Use // comments througout, improve comments style, move opening braces to function definition lines.
-
- 25 Dec, 2017 5 commits
-
-
Mark Adler authored
-
Mark Adler authored
-
Mark Adler authored
-
Mark Adler authored
This also cleans up the termination of the read thread for those cases where the entire input was not read, but the read thread is no longer needed.
-
Mark Adler authored
This allows for weak linking with a zlib library whose version is older than the zlib header file used to compile pigz. The version is checked at run time, and deflatePending() is not called if the linked library does not have it. This commit adds a check as well when invoked for the earliest acceptable version of zlib (1.2.3).
-
- 14 Dec, 2016 1 commit
-
-
Mark Adler authored
BSD make doesn't understand $^, and the Sun make has a problem with $<. This avoids the automatic variables by explicitly specifying the objects and targets. This also avoids a compiler complaint for a #error directive by putting the message in quotes.
-
- 13 Dec, 2016 1 commit
-
-
Mark Adler authored
Facilitate compiling pigz without Zopfli for implementations that do not need it and that desire to minimize the code size.
-
- 12 Dec, 2016 4 commits
-
-
Mark Adler authored
-
Mark Adler authored
This replaces the previous -M for --maxsplits with -J. (Hopefully no one was using -M. Probably not.) This fills a hole left by the change of -n, --no-name to also not store/restore the modification time. Now -nM will store/restore the modification time, but not the name. The full set of combinations is -n for neither, -N for both, -nM for just the time, and -Nm for just the name. Note that the order is important. -n or -N will overwrite the effect of a preceding -m or -M. The default for compression is -N, and the default for decompression is -n. So -N or -n can be left off for those cases, respectively.
-
Mark Adler authored
Though this is not consistent with the name of the option, this commit makes pigz compatible with gzip's use of -n and --no-time.
-
Mark Adler authored
Apparently gzip has supported -m all along to not include the modify time, but it wasn't in the command help list. This commit adds -m as an alias for --no-time. -T is retained as an alias for --no-time for backwards compatibility with previous versions of pigz.
-
- 15 Oct, 2016 3 commits
-
-
Mark Adler authored
-
Mark Adler authored
-
Mark Adler authored
-
- 12 Oct, 2016 1 commit
-
-
Mark Adler authored
Microsoft has not implemented and refuses to implement C99 aspects that are not also part of C++11 or later C++ standards. (Note that the "99" in C99 refers to 1999. That was the year my younger son was born, who is now applying for college.)
-
- 11 Oct, 2016 1 commit
-
-
Mark Adler authored
Before -K would fail when the compressed or uncompressed data, or the offset of the central directory, was 4 GiB or more. Now -K will use Zip64 data descriptors in the local and central headers, and will use the Zip64 end records as needed.
-
- 02 Oct, 2016 5 commits
-
-
Mark Adler authored
-
Mark Adler authored
-
Mark Adler authored
-
Mark Adler authored
pigz currently will not generate a ZIP64 format .zip file if the input uncompressed or compressed lengths are greater than or equal to 4 GiB. This commit issues a warning in that instance that the resulting .zip file will be unusable.
-
Mark Adler authored
This off-by-one error resulted in useless reexamination of the same files by descending twice into each directory.
-
- 27 Sep, 2016 1 commit
-
-
Mark Adler authored
-
- 16 Aug, 2016 1 commit
-
-
Mark Adler authored
-
- 30 May, 2016 7 commits
-
-
Mark Adler authored
-
Mark Adler authored
-
Mark Adler authored
-
Mark Adler authored
-
Mark Adler authored
The consistency of block sync markers depends on zlib 1.2.6 or later. pigz -vV will display the zlib version linked with pigz.
-
Mark Adler authored
This forces two sync markers between independent blocks, giving a nine-byte signature 00 00 ff ff 00 00 00 ff ff to search for when processing compressed data in parallel.
-
Mark Adler authored
-
- 20 Aug, 2015 1 commit
-
-
Mark Adler authored
-