The .tar plug-in should remember the offset of the file

Roland Illig · Post by **Roland Illig** » 06 Mar 2008, 17:35

I have made some backups using an uncompressed .tar file. When I want to extract a single file, the Salamander presents me with a long, long list of files that are skipped. It would be much more efficient to remember the offset where the file was found in the .tar archive and then to extract the data from that position in the file.

I also have the impression that the .tar plugin doesn't make use of seeking in a file. I have another archive with two entries, both about 300 MB large. When I open the archive from a CD-ROM, it takes a very long time to show me the listing. Much more time than I expect for a simple sequence of OpenFile(), ReadFile(header1), SetFilePosition(header1->size), ReadFile(header2), CloseFile().

Roland

Post by **Tomas Kopal** » 10 Mar 2008, 09:57

Yes, there is a lot of room for improvements in the tar plugin. It was written in the simplest possible way just to get something started. We plan to get back to it in the future and improve it further.
If you have any other ideas what can be improved, post them here, so we don't forget to include anything when we start working on the plugin again.

SvA · Post by **SvA** » 10 Mar 2008, 22:04

I downloaded the CSS 2.1 documentation the other day. The file is named css2.tgz.gz (A strange name, I know. It's actually a gzipped tar. The additional .gz had been added by the Safari (Windows) download manager based on the mime type).
When I unpack it, the suggested directory name is css2.tgz. It should rather be css2. After all the plugin processes gz as well as tar (which I like). (Btw, if the name was css2.tar.gz, Salamander would handle it just the same.)