Digital Mages

Digital Mages

Jan 10, 2024


Parse::PlainConfig has been updated to fix a small problem with the handling of array/hash prototyped parameters that had no contents.

Sep 23, 2023


Parse::PlainConfig has a new release whose main feature is the ability to subclass your config modules. Each subclass will automatically incorporate the parent classes' parameters, prototypes, and defaults.

It also has minor doc updates, plus debug output reworked to take advantage of Paranoid v2.10's enhancements.

Mar 8, 2022


Paranoid has a new release that adds some of the most visible changes we've had in awhile, with a refinement in the Paranoid::Debug's syntax for functions and methods, along with a new rendering mode for its output. All of the Paranoid modules have been modified to use the new subPreamble/subPostamble functions. It not only reduces the amount of code needed for internal diagnostic output support, but it now has better summarization features for various data types. For I/O functions that reported on arbitrarily sized arguments being passed, this reduces the noise quite abit.

Paranoid::Data::AVLTree also got its promised save to/load from a file, along with some basic profiling support. This makes this module a bit more usable for those programs fine with working from memory data sets, but needing a way to preserve that data between executions. While one could use Data::Dumper for this, having a direct support and a dedicated format allows us the ability to spool the tree to disk in a manner that minimizes tree rebalancing operations on loading, as well as not having to read a huge string into memory that needs to be evaluated. There should be some performance gains if you need to do this frequently.

Paranoid::Data also inherited the functions I wrote for the FileMultiplexer to accomodate 32 and 64-bit file system support, with quads2Longs and longs2Quad conversion functions. The impetus for this was essentially my attempt to transparently support both scales of systems with one record format. We used the conversion functions to let us store 32 or 64 bit values in one field by always allocating 64 bits. On smaller scale systems, only 32 bit values are effectively supported, but the file can be safely interchanged with 64 bit platforms with no modifications. It's only when a 64 bit system working on the file actually overflows into the upper 32 bits that the file becomes unusable on 32 bit platforms. And the code conversion functions check for those situations, and error out when it sees a value it can't natively support. End sum, you have file portability across both scales while the content doesn't require 64 bit values, but you have the ability to scale on supported systems when you do.

For an example of how this works in practice, you can look at both Paranoid::Data::AVLTree and Paranoid::IO::FileMultiplexer to see that concept in action.

Finally, along with some documentation cleanup in other modules, I also changed the default PATH used by pSecureEnv to include the core sbin directories. I really can't explain why I didn't do that in the first place, but it's there now.

But, wait, that's not all! I also did a minor transformation on Paranoid::BerkeleyDB to leverage the new Debug support. Admittedly, though, the true driver was to fix a bad test in the test suite, that made the code base to appear to fail a test when it really didn't. There's some serious egg on my face for not catching that my last published tarball had that error for as long as it did. We'll never mention this again, though. ;-)

Jan 20, 2022


One of the things I've always done poorly with is promoting the use of my own modules. This is primarily due to the fact that I wrote these to make my life easier, any benefit to others is entirely accidental. That said, I do believe that what I've built over the last decade plus does have merit. If not the code itself, the intent behind it. For that reason, I am expanding the sections on this site to include educational information. The Paranoid modules being my first topic of discusion.

If you have any interest, at all, from a developer's perspective, you can peruse those at Introduction to the Paranoid Framework for Perl.

Dec 28, 2021


Ah, the holidays. That time of year when I look at my personal projects and say "WTH!" Did I really spend the whole bloody year doing... work?! Well, not entirely just on my professional responsibilities, but damned near. Almost forgot to finish up a work in progress and get it out the door. And there began the mad, holiday rush...

Paranoid has been updated to include several minor improvements, but also two major pieces of functionality: a new flock stack that better supports transactional I/O patterns, along with a file "multiplexer".

The flock stack is a global switch that affects all I/O functions provided by Paranoid. It's not enabled by default right now, but I can see that changing in the next major release. The point of the lock stack is to track flocks on files and prevent them from being degraded while being part of a larger transaction. This should make it tremendously easier to generalize code blocks while increasing code useablility. There is a much better write-up of a typical use-case scenario in the Paranoid::IO man page to provide more details.

The file multiplexer makes heavy use of the lock stack, to provide concurrent multi-process access to multiple data streams embedded within a single file. At its simplest, the "mulitplexer" can serve as a basic archive format. At its most complex, it can provide a database backend file, similar to sqlite or BerkeleyDB. If you wondered why I had previously added an AVL binary tree class, now you know where this is going.

If I take as long for the next update, call in the hit. I deserve it.

Dec 31, 2020


Paranoid has been updated to include some new functionality to Paranoid::Process, as well as the addition of an AVL-balanced tree implementation in Paranoid::Data::AVLTree.

This also includes updated licensing information to reflect the acquisition of the Paranoid trademark by the Pleasant Solutions Group.

May 24, 2019


Class::EHierarchy has been updated to account for a bug in the conceive class method, in which the child object will be left alive even if it fails initialization. This has been fixed.

May 1, 2019


Misc Network Tools is a set of miscellaneous tools I've had knocking around for over twelve years. With this release, though, I've added a few new ones that might be useful to some of you dealing with large sets of IP data, like blacklists or what have you.

ipconsolidate is a handy script that can take any kind of arbitrary text and extract all of the IPv4/IPv6 IPs and network addresses, eliminate redundant entries, and print a sorted list. What makes this a little special is that if, say, you have both 192.168.0.54 and 192.168.0.0/24 in the list, it will eliminate the single IP because it knows it's already part of the network range. It can also do the same for 10.0.200.0/24 and 10.0.0.0/8, as well.

ipgrep is a similarly capable grep. It can find lines in any input that logically match the address you're searching for. It will print lines that contain CIDR ranges that would encompass the IP you're looking for or, conversely, print all the lines with IPs that would be part of the CIDR range you're looking for. It also supports exact and quiet matching just like traditional grep.

Apr 17, 2019


This isn't just another Paranoid update this time. This time we have some significant enhancements of various APIs which continue to save me time on some of the more repetitive tasks.

Paranoid::Args, for example, has reduced the number of options required for option templates, with some being set automatically when obvious semantics are identified. On top of that, some of the more standard options now have importable templates.

Paranoid::Debug now provides constants for debug levels 1 through 8 which are intended for developer use and won't collide with Paranoid's internal debugging.

Paranoid::Filesystem has improved pmkdir's handling of permissions more predictable, and ptranslatePerms now allows you to pass translated and untranslated permissions with impunity, while still giving you the expected form out.

Paranoid::Log, however, has the most significant work done. It now supports both STDOUT and STDERR, which doesn't at first blush seem like much, but with a new plverbosity function, you can now map PL_DEBUG - PL_NOTICE log messages to STDOUT, and PL_WARN - PL_EMERG to STDERR with one call. This gives you a way to tune end-user verbosity for console messages without having to subject them to pdebug diagnostic input.

But, wait, there's more! All too often I've had to lace my code with dual calls to plog and pdebug. There's now a PDebug logging mechanism that can eliminate that redundancy, distributing both diagnostic messages to logs and STDERR as needed.

That brings us to one more significant enhancement. pdebug has long supported printf functionality with automatic array/scalar coercion and undef translation to strings. As long as you're enabling the PDebug mechanism, you can now leverage those same capabilities in plog, regardless of what logging mechanism those messages are distributed to.

Finally, like always, there are other small tweaks and improvements, but you'll need to peruse the changelog to get the full picture.

Aug 8, 2018


Another day, another Paranoid update. Mostly minor updates, one significant improvement (pfork()), and one embarrassing bug (pflock()). A better test would have shown how it wasn't working quite the way it should, so now we have that better test.

It is what it is, but, that said, I do have to say the rewrite is still paying dividends in convenience of use over the old rev. We might actually be getting somewhere with this. ;-)

May 3, 2018


Leif Sawyer pointed out a wee bug in the logic for NS server counts when the tool isn't run in strict mode. I also fleshed out more of the documentation. So, go get the latest zone-lint.

May 3, 2018


Life really gets in the way, sometimes. That said, I've found some time to brush some dust off an old script, and it may have some usefulness for you, as well.

zone-lint is a DNS zone lint check tool which provides some RFC-based validation of published DNS records. I've found it good practice to run this script after every update. I originally wrote this over ten years ago, but have now ported it to the latest Paranoid API, along with adding IPv6 support.

Apr 26, 2017


I couldn't very well leave Net::ICAP in a completely broken state while working on my applications, so I banged out a quick port, along fixing a few outstanding bugs reported by the good folks on rt.cpan.org. This module more than likely needs a complete review and rewrite, but one thing at a time.

Mar 23, 2017


Aaannnnd, just like that, here's another update. SREZIC@cpan.org caught my oversight in Parse::PlainConfig where I forgot to update the minimum required version in Makefile.PL for Class::EHierarchy. So, for that reason, P::PC gets a second release today.

Mar 23, 2017


I finally finished porting my dependent modules to the new Class::EHierarchy API. During the course of the ports I found a few minor bugs in Paranoid, as well.

All that said, I will have to admit some to small API alterations to Paranoid::BerkeleyDB, which is probaby bad form. I don't believe this module is widely used, however, so I'd rather make it now than wait. It does make the module behavior more consistent.

So, without further adieu, the following new modules have been released both here and on CPAN:

Jan 23, 2017


A quarter of a year is a long time to disappear. The last quarter of any fiscal year is always a killer as projects have to be wrapped up before the new budgets come in. Then, there's the holidays and a general desire to enjoy the real world with the family rather than stay mired in the digital sphere.

Regardless, I'm back. And I have been generally running amuck generating all kinds of extra work for myself. While working on autofwd, which I'm quite excited about, I happened to have snagged myself on several corner cases where the purported promises of Class::EHierarchy weren't quite delivered. The deeper I digged, the more it became apparent that I had really done a rubbish job of justifying that module's existence.

The only sane thing to do was to throw the entire code base out and start from scratch. I am rather ashamed of what ridiculous spaghetti code it was, I'm not quite sure how I let that malformed beast escape the lab. What was needed was a very methodical object build and destruction process, along with a much more thorough testing suite.

That said, I have good news and I have bad news. The good news is that Class::EHierarchy is complete. The bad news is that you can't have it, yet. Like the rewrite of Paranoid, the rewrite required an API break. Unfortunately for me, that also means that two of my more newer code updates now need to be updated to accomodate that.

End sum, because I don't want any psuedo-stable modules on CPAN broken I need to update them in order to do a coordinated release that keeps everything working.

Stay with me, the improved API(s) are worth it, and will benefit code quality built off of it. Thank you for your patience.

Sep 19, 2016


Small update to Paranoid for added functionality to Paranoid::Process. I added a signal dispatcher so I wouldn't have to worry about writing monolithic signal handlers. Now I can just add subroutines to a queue and have the dispatcher process them all as signals are recieved.

Also added minor updates to Microscopic Certificate Authorithy. No show-stoppers in terms of bugs, just some documentation updates and a correction to the logging mechanism.

Sep 11, 2016


Many tangents later it became apparent that others might get some use from a little package I've been using for years. Unfortunately, that meant refactoring the interface a wee bit and documenting it. Long story short, Microscopic Certificate Authorithy is now available.

Nothing too fancy, it's just an extremely simple way to get an internal CA up and running so you can quit adding a hundred exceptions in every browser you use. So simple that it only needs OpenSSL and GNU make in a UNIX environment to work.

And yes, I will get back to work on finishing my ports... :-P

August 6, 2016


The best laid plans of mice & men go oft' astray... Minor update this time for Parse::PlainConfig. Debian folks found a typo and I found a bug with the error reporting in the parser.

The Debian folks also find a trove of typos in Paranoid. Another gentleman (theodore@grcrun11.gr) also pointed out the need for an improvement to the IPv6 regex, so that's there as well.

June 23, 2016


A few bug fixes and some expanded documentation necessitated the release of Paranoid v2.01. CPAN should be updated before long.

June 21, 2016


Not to sound like a broken record but I just uploaded Paranoid::BerkeleyDB to CPAN and this site. CPAN Testers pointed out a wee oversight of mine in which I was relying on a BDB 4.6+ call that failed on dists installed with older versions. Added some intelligence to fix that.

June 17, 2016


Just uploaded Paranoid::BerkeleyDB to CPAN and this site. Minor update, mostly documentation updates some small code adjustments.

June 13, 2016


Whew! Missed that by a mile, eh? So, I obviously didn't finish by the proposed deadline, but it is still coming. I've got a minor update to Paranoid::BerkeleyDB and I believe I will have everything in place to finish the port of autofwd.

This port isn't just a port, however, which is one reason why it's taking so long. It's essentially a complete rewrite meant to make it even easier to administer and far more flexible as well. With the help of Sander Klein as a sounding board I think this is going to be an update more than worth your while. It really can lend itself to so much more than just automated firewalling, but we'll see what people come up for it.

So, stay tuned, and thank you for your patience.

May 26, 2016


One of the joys of porting apps to the new API is in finding the little things we overlooked that make a big difference in usability. Found one such animal in Parse::PlainConfig in regards to a new feature of the parser: prototypes. As I'm reworking autofwd to leverage it it became obvious I was lacking an accompanying feature to make it really useful. So, now that's done. The update to the parser is live on this website now, and should be on CPAN within the day.

May 23, 2016


I'm not really trying to blow my deadlines, but I figured, hey, here's a program that's already using the new API. You may find it moderately useful: yapw.

May 19, 2016


Parse::PlainConfig is also live. This is updated to not only use the new Paranoid API but is a complete rethink on config parsers. This is tremendously nicer to use as a developer, but it does give up generation capabilities and the ability to parse arbitrary, free form configs. This now assumes that you have a well defined set of configuration parameters that are strongly typed by default. Much better suited for all of my use-case scenarios, at least. YMMV.

May 18, 2016


Paranoid v2.00 is now live, It should be up on CPAN soon as well. Please be aware that with this update I've also split out three modules into their own distributions since they required optional dependencies. This removes that ambiguity:

May 13, 2016


After quite the hiatus from posting it's time to get this site up to date. While I loathe updating web sites it's not like I haven't been coding, I've just haven't been sharing. If there ever was a time to share, however, it would be now.

I've spend most of the last ten years eating my own dog food, building most of my internal (and professional) code bases on my libraries. While those libraries have been effective, they haven't always been pleasant. As I've evolved as a hacker I've had to resign myself to dealing with a pragmatic reality that has refused, in many cases, to bend to my will.

Any of you who know me personally -- apologies for that, btw -- know that I've had a very long fixation on secure programming. All effective security has a cost and finding the sweet spot between effective security and usability is a constant balancing act. Both for the user and the developer.

I have been guilty for a long time of erroring on the side of the user. Not so much in usability -- again, sorry for that ;-) -- but in trying to guarantee effective security. It is in that obsessiveness that I managed to cross both the users and the developers. The best security in the world means nothing if it's so unpleasant that no one wants to use it.

All of which brings me to one of the most important lessons I could learn as a library developer: Don't be a dick, especially to other developers. It is perhaps ironic that as a cyber-recluse it took some time for me to recognize while slinging code that "that Paranoid guy's a dick." When I don't like using my own code, there's a problem. Therefore, I must reform.

I have spent considerable time over the last year rewriting my libraries, sometimes rewriting them entirely from scratch. The intent was to make the API more enjoyable to use while minimizing any adverse impact on the security goals.

For my Perl modules most of this is in the little details. Eliminating the use of function prototypes to make it easier to pass in arguments in list form, rather than explicitly by value. Or deploying prototypes only where it makes syntax more pleasant so the developer doesn't have to explicitly pass data structures by reference. By not croaking at every imagined offense, instead just exiting with a false return value.

With some of my foundational libraries already updated the race is now on to port the rest of the software that relies on it to the new API. That does mean, unfortunately, that software on this site is effectively broken if you grab the last revs of everything and try to build with it. My commitment to you is that I finish porting the broken software to the new API by the end of the month.

So there we have it. The ogre pokes his head out of his cave and has makes his proclamation. I have no intention of stopping my existence as an ogre, but in a small deviation I am very interested in feedback to these updates as I upload them. I am very excited about these changes, it's been a revelation -- to me, at least -- how much an API can affect one's eagerness to sling new code. The old APIs had really worn me down.

I'll leave it at that. Thank you for your time. Happy hacking!