Posts for October 2016

2016-10-10: rra-c-util 6.1

This is my collection of supporting libraries, Autoconf macros, and test code for various projects.

This release fixes return-value checks for snprintf to avoid a few off-by-one errors (none of which should have been exploitable, but better to be safe and correct). It adds a new RRA_PROG_CC_FLAG macro to test compiler support for a specific flag and a new RRA_PROG_CC_WARNINGS_FLAGS macro to probe for all the flags I use as my standard make warnings target. And it fixes some problems with one utility due to the removal of the current directory from @INC in the latest Perl release.

You can get the latest version from the rra-c-util distribution page.

2016-10-10: remctl 3.13

remctl is a client and server that forms a very simple remote RPC system, normally authenticated with Kerberos, although including a remctl-shell variant that works over ssh.

This release adds forced-command support for remctl-shell, which allows it to work without enabling setting environment variables in authorized_keys. This may be a preferrable configuration to using it as an actual shell.

Also in this release, the summary configuration option is allowed for commands with subcommands other than ALL, which allows proper generation of command summaries even for users who only have access to a few subcommands of a command. It also adds some build system support for building binaries with -fPIE.

You can get the latest release from the remctl distribution page.

2016-10-17: pgpcontrol 2.5

pgpcontrol is the collection of the original signing and verification scripts that David Lawrence wrote (in Perl) for verification of Usenet control messages. I took over maintenance of it, with a few other things, but haven't really done much with it. It would benefit a lot from an overhaul of both the documentation and the code, and turning it into a more normal Perl module and supporting scripts.

This release is none of those things. It's just pure housekeeping, picking up changes made by other people (mostly Julien √ČLIE) to the copies of the scripts in INN and making a few minor URL tweaks. But I figured I may as well, rather than distribute old versions of the scripts.

You can tell how little I've done with this stuff by noting that they don't even have a distribution page on my web site. The canonical distribution site is ftp.isc.org, although I'm not sure if that site will pick up the new release. (This relies on a chain of rsync commands that have been moved multiple times since the last time I pushed the release button, and I suspect that has broken.) I'll ping someone about possibly fixing that; in the meantime, you can find the files on archives.eyrie.org.

2016-10-26: DocKnot 1.00

I'm a bit of a perfectionist about package documentation, and I'm also a huge fan of consistency. As I've slowly accumulated more open source software packages (alas, fewer new ones these days since I have less day-job time to work on them), I've developed a standard format for package documentation files, particularly the README in the package and the web pages I publish. I've iterated on these, tweaking them and messing with them, trying to incorporate all my accumulated wisdom about what information people need.

Unfortunately, this gets very tedious, since I have dozens of packages at this point and rarely want to go through the effort of changing every one of them every time I come up with a better way of phrasing something or change some aspect of my standard package build configuration. I also have the same information duplicated in multiple locations (the README and the web page for the package). And there's a lot of boilerplate that's common for all of my packages that I don't want to keep copying (or changing when I do things like change all eyrie.org URLs to HTTPS).

About three years ago, I started seriously brainstorming ways of automating this process. I made a start on it during one self-directed day at my old job at Stanford, but only got it far enough to generate a few basic files. Since then, I keep thinking about it, keep wishing I had it, and keep not putting the work into finishing it.

During this vacation, after a week and a half of relaxing and reading, I finally felt like doing a larger development project and finally started working on this for long enough to build up some momentum. Two days later, and this is finally ready for an initial release.

DocKnot uses metadata (which I'm putting in docs/metadata) that's mostly JSON plus some documentation fragments and generates README, the web page for the package (in thread, the macro language I use for all my web pages), and (the other thing I've wanted to do and didn't want to tackle without this tool) README.md, a Markdown version of README that will look nice on GitHub.

The templates that come with the package are all rather specific to me, particularly the thread template which would be unusable by anyone else. I have no idea if anyone else will want to use this package (and right now the metadata format is entirely undocumented). But since it's a shame to not release things as free software, and since I suspect I may need to upload it to Debian since, technically, this tool is required to "build" the README file distributed with my packages, here it is. I've also uploaded it to CPAN (it's my first experiment with the App::* namespace for things that aren't really meant to be used as a standalone Perl module).

You can get the latest version from the DocKnot distribution page (which is indeed generated with DocKnot). Also now generated with DocKnot are the rra-c-util and C TAP Harness distribution pages. Let me know if you see anything weird; there are doubtless still a few bugs.

2016-10-28: Term::ANSIColor 4.06

A small maintenance release to my Perl module for generating and manipulating ANSI color escape sequences.

In 4.00, I added 256-color support, using ansi0 through ansi15 for the colors that match the normal 16 colors and then rgbNNN and greyN names for the RGB and greyscale colors. One module user requested the ability to address all of the colors via ansi0 through ansi255 so that there's a consistent naming scheme for all the colors. This release adds that. (The more specific names are still returned by uncolor() when reversing escape sequences.)

This module tends to get included all over the place, so I did spend a bit of time when preparing this release trying to determine if adding a bunch more entries to various internal hash tables and arrays would noticably increase memory usage, since Perl is notoriously bad about memory consumption. It does cause a small increase, but it's on the order of about 100KB, and a minimum Perl program to load the module requires about 5.5MB of memory (aie), so it wasn't enough for me to do anything about.

It does look like if I lazily added entries to the built-in hash tables and instead added some more code to calculate escape sequences on the fly, I could save about 300KB of memory usage in the module. Not sure if it's worth it given how small the memory usage is compared to Perl itself, but maybe I'll look at that later when I'm feeling like fiddling with the module again.

(Oh, and all the documentation was regenerated by DocKnot, since I'm still having fun with that. It needed a few new features, which will be in an upcoming 1.01 release.)

You can get the latest version from the Term::ANSIColor distribution page.

Last modified and spun 2017-03-25