Upgrade from 5.20.2 to 5.26 lost several Win32 packages, even though I saved and restored Per62\site. Now I find that, to reinstall those modules, I need to learn all about “projects” and their “virtual runtime environments.” I’m using Perl to write small scripts for managing downloads and audio files on my system. Using Perl is hard enough without having more obstacles placed in my path.
UPDATE: 5.26 doesn’t even uninstall; it complains about files being in use. It’s necessary to manually delete all files in C:\Perl64 and then run the uninstall again to remove it from the list of installed programs.
Hi @Shinelikethunder - I’m sorry that you’re having issues upgrading to 5.26 - I’d be happy to take a crack at creating a Perl 5.32 runtime for you that you could install via the command line. Do you have a list of packages that you know you need supported?
Cheers.
1 Like
Hi,
Thank you very much for your kind offer. Sorry for the delay; I’ve been busy with other things. The “panic: swash_fetch” problem with an RE search for text in binary data that led me to attempt an upgrade from 5.20.2 has been solved. I was using:
open($fh, “<$filepath”) or die “Couldn’t open $filepath\n”;
binmode $fh, “:raw”;
to open a uTorrent file resume.dat for reading. When I changed that to:
open($fh, “<:raw”, $filepath) or die “Couldn’t open $filepath\n”;
the problem went away. Apparently, Perl expects strings read from the file in the 1st case to already be in its internal encoding. (I believe that this is a bug; perhaps you could look into whether it has since been fixed or why strings read from a file are apparently marked as binary in the 2nd case, but not the 1st.)
(I had no luck whatsoever with the Bencode module that is supposed to be able to decode uTorrent files. It barfed on every single bencoded file in the uTorrent folder on two systems.)
Most of the installation problems I had with 5.24 and 5.26 were related to the Encode module, specifically Encode::Detect::Detector and Encode::Detect::Detective. If you have a new version that supports those modules, I’d be happy to try it. I’m reading CUE files that are supposed to be in UTF8, but apparently the standard isn’t enforced, so I need a way to detect the encoding used. If you’re aware of any better ways to test the encoding of a text file, I’d appreciate suggestions. (Because all Western charsets will yield “valid” decodings of an encoding made by any of the others, I’d really like to be able to determine the most likely encoding using a dictionary search of the words that differ between two decodings of the raw bytes, but I doubt that such a thing exists.)
Thanks,
Loren
Hey @Shinelikethunder - do you have a list of modules that you want included in the project?
I created a runtime with the two you mentioned:
You can make your own copy of it by clicking the “Fork It” button in the header
You can install it via the command line and then use the state tool to add additional packages. Here’s a cheat sheet of State Tool commands…
State Tool aims to replace PPM, which we cannot maintain for a variety of technical reasons. You can also add packages via the platform UI on the Configuration tab and then run state pull to update your local installation.
Let me know how that goes or if you run into trouble!