Well I got through the problems with Apache2::Module::get_config. Aparently the object which comes as the first argument to the method for the custom config directive is not where the config parameters should be stored. Looking back at the documentation http://perl.apache.org/docs/2.0/api/Apache2/Module.html, I can see that it is clearly shown that $srv_config is being fetched via a call to get_config from within the custom directive subroutine, and that apparently is where the config should be stored... but it does seem kind of counter intuitive to me. But no use second guessing that now, at least I got it.
On a positive note: before I discovered my mistake with get_config, I thought perhaps it was getting mangled by Apache2::compat, so I spent much of the day finding and jumping through hoops to get everything to run without that. Though that effort didn't fix the problem I was chasing, it will be better in the long run to not need Apache2::compat, as that would drag down response times.
In my ongoing efforts to port slashdot onto modern revisions of Apache and ModPerl, I've made significant strides. This weekend I put in a few extras hours of work, just because it was on my mind.
Currently I am banging my head against Apache2::Module::get_config. I thought I had appropriately adjusted the code to use the new simplified mechanism for defining and handling custom apache config directives. Having added plenty of "print STDERR" debuging into my modules, I can see that the new directives are being read, and the associated sub routines to handle them are being called at server start time. However, calls to get_config from inside a request handler are always returning either 1 or undef. As far as I can tell from reading the documentation, it should be working. As I type this, it occurs to me that it could be getting fubared by Apache2::compat...
Off to test that theory!
As my colleagues are working to improve the comment system, I'm thinking about it. I had an idea today, and I'd love to hear what anybody reading my journal thinks about it:
Would it be useful to have a control that would either expand or collapse all of a comments ancestors, descendants, or siblings?
So, what do you think?
I am continuing forward with the re-platforming effort to move to Apache2.2 and ModPerl2. I have in the last few months be doing little bits here and there where I could squeeze it in between other projects, but today I got a good half day put into it. Before today I had updated the Makefile.PL's where appropriate to move away from Apache::ExtUtils, and instead use ModPerl::MM. I spent much of today finding all the CPAN modules that we rely on, and getting them built under the proper version of perl. After that I spent an hour or so tweaking config files where needed for the newer version of apache. Before heading home for the day I was actually to the point of trying to start Apache. It almost started, but I apparently missed something in updating the system to use Apache2::Module::add to handle the custom Apache config directives. But that will have to wait for tomorrow.
In comment system news: my colleagues are currently working on mock-ups for a few new controls to allow greater control over how many comments get loaded on an article page. Also trying to brainstorm around an easier/more intuitive way to handle setting thresholds for comment visibility (the current implementation is the "slider" widget with the two handles at the right side of the comment section header).
And there were, and it was good, for a while.
The comment system, and the discussions that take place in it, are a very important part of what makes slashdot what it is. I think it is a very powerful system, into which lots of time, thought, tests, trial and error, blood, sweat, unicorn meat, and tears have been put over the years. It is currently very complicated, somewhat fragmented and not well understood my a large number of readers. As I mentioned in a reply in one of my previous journal entries, the comment system is due to receive some much needed attention in the coming weeks. To that end, I am pleased to note that a release will be coming out today to address an issue we've found in the dispersal of moderation points. More than one or two readers have asked if there was something broken, or if they had somehow lost "moderation privilege", and I do believe that starting tomorrow, that issue should be resolved (thanks to the hard work of my friend/colleague vroom.)
Stay tuned for news of more love to come. (And feel free to let me know if you have special requests or concerns about the comment system)
Well the Hall of fame seems to be working for the few features it offered initially, and a few people have given me some good feedback and suggestions for features it could/should have. Once its been live as is for some period, I'll put my focus back on it, and hopefully implement some of the suggestions I've gotten.
In the mean time, I will be spending some time on back end work. While I may not be able to point to any "shinny new" stuff, I hope readers will see the results of this work in the form of stability, speed, and reliability. I am also looking for a contractor to help me out for a few months with the platform migration to Apache 2 and mod_perl 2... hence the title of this post. If you have a strong background with LAMP stack applications, and think you have what it takes to help us improve slashdot, I am accepting resume's. The position would be short term (2-3 months), but it is a chance to get into slashdot, and show us what you've got.
So, Can you ? (leave me a comment if interested and I'll get you contact info)
P.S. there is also a full time position open for an entry level/associate engineer on my team. Go Here for details
Well, the new Hall of Fame is now live and open for business. I am trying to be patient, and to wait for it to live in the world for a bit before I jump into to look at the stats. Mostly I just hope that at least some people find it interesting, or helpful. I am moving into my next project next week, but will turn my attention back to the HoF from time to time, and hopefully add features. Please feel free to let me know if the new HoF is a disappointment, or if it is great, if you think it could be better if...
I am excited for my next project, though it will be a lot of work... Migrating Slashdot from Apache 1.3 and mod_perl 1 to Apache 2.2 and mod_perl 2. Though I have undertaken this sort of project of a few large systems, Slashdot will have some new puzzles to solve. I hope in the course of this project I will also be able to hammer out a few persistent bugs, and probably do some stream lining to get the page render times down a bit.
With the release of the new "Hall of Fame" comming later this week, I've squeezed one more feature into it for the first release. We will have a "Staff Favorites" section. Originally, I was thinking this would be only "Editors Favorites", but I thought (and my colleagues seem to agree) that it would be nice to show the "Favorites" of a larger portion of our team. That being said, I am not compelling any one (not even the editors) to share their "favorite" stories, but I will be giving the option to all of our engineers in addition to the editors. If that picks up, perhaps others will want to join in.
I wonder... if we offered the feature: would our readers like to be able to mark their own favorite stories? Maybe their favorite comments as well? This information could, if a user wanted it to, be displayed on their user page. Perhaps also, if we biuld out more of the system's "friend or foe" system, users could then browse their friends favorite stories/comments/submissions? (Must be honest: part of this htought is inspired by my friend/colleague Cogent's ideas.
Wow. My last journal entry really droned on to long. Feel free to tell me to shut up next time.
First round draft of the new "Periodic" Hall of Fame (as I've been calling it) should be showing up in the wild in this comming week. For the first release it is really just a sub set of the existing hall of fame, though broken down to smaller chunks of time. The really new piece here, however, will be the display of a the current users own stats next to each category. So next to "Most Active Submitters" a logged in user will see "My submissions", and next to "Most active Commenters" they will see "My comments". I hope by doing this people will be able to see that the new comers are not really that different from the grey hairs on a smaller time scale.
One other quick note: A feature we may be able to bring in V2 might be a visual representation of some of the data. For example perhaps next to "My comments" there may be a graph of # of comments a user has left per week in the last month. What do you think: Would that be cool? Usefull? Overkill? Please, share your thoughts...
As I've been working this week to build a supplement for our existing hall of fame, I've been making decisions (or guesses) about many things, some the I expected, and some that I didn't... We will be releasing this thing into the wild soon, and some of the decisions I've made may be reversed once we see how readers actually interact with and re-act to it. All of you that are smart/lucky/bored/crazy/unfortunate (you pick the adjective) enough to be reading and following my thoughts are invited to leave me you're thoughts here.
Our current "Hall of Fame" has two parts... "All Time" and "Past Year". "All Time" is self explanatory. "This Year" could be interpreted in two ways: during the current calendar year or during the last roughly 365 days (current hall of fame uses the latter). For the new bits of the Hall of Fame, I will be breaking it down in "Weekly", "Monthly" and "Quarterly". Each of these increments of time are open for the same interpretations as "This Year", and by nature of being smaller in scale, the differences can have a much more immediate effect, and both interpretations have both strength, and weakness as I see it.
My first instinct is/was to use absolute measurements for these (weekly is Sunday-Saturday, Monthly is from the 1st on, Quarterly is January - March etc.). The strength I see in this would be that we have a clearly defined range of time, with a beginning and an end. I think that could encourage new comers to the site to engage more immediately (at least at the weekly level) as at the beginning of the time span everyone, old and new, quiet and reserved or loud and obnoxious, would start at 0. The other possible strength here would be that with a define beginning and end we can mark the end of a time period and record the winners. The weakness in this approach (which I also argue is a strength) is that at the beginning of a time span the "Hall of Fame" would be full of low scores and therefore not particularly useful until we are a significant portion of the way into it. The (perhaps less obvious) problem created here would be that the time when this has the most complete (and arguably) the most useful data is near the end of the period and by the time the data set is complete, we have moved into the next period and started all over again.
Currently I am leaning toward the "rolling window" approach, so "Weekly" will always have the last 7 days of data. This implementation is easier to code because I never need to worry about how far into a week, and month or a quarter I am. It will also mean that regardless of when a reader looks at it, it has useful data, even if it is on the first day of the month or week. The first drawback would be that with no beginning and no end, on race ever ends, and no one ever gets to claim victory. The second would be that new comers will always see the bottom of the "Hall of Fame" as where they would need to be at the end of a week or month, which could seem unreachable.
The other issue I find myself grappling with on this topic is one of quality. We all know that there is more to being a part of something than just showing up. Slashdot is a community created/supported site, and we depending on community contribution not only for submitting stories, but also for participating in discussions, and even helping to moderate the discussions. Sadly, we also all know that not all contributions are positive. Everyone can, and lots of people do, submit stories, and make comments. The bulk of these are good, but here, as everywhere, some are just not. When we recognize members of our community we want to recognize the quality, not just the quantity, of their contributions. So for example we want to recognize those who submit stories to us on a regular basis, but we do not want to recognize the "entities" that are flooding our inbox's with spam. Similarly we want to recognize those who participate in discussions, but probably not the clowns who spends time trying to hit the comment button the first just so they are can say "Hey! Look at me, I'm first!". This question can be answered in many ways, none of which will be perfect, and most of which will piss somebody off.
Another dilemma comes into play in this decision... we want people to contribute, and we want to recognize those who do so. It seems only fair to have a well defined set of criteria on which to select who we should recognize. The problem is deciding not only what criteria are fair, but ideally we like to prevent people from "gaming the system". I'd love to live in a world in which no one would do that, but the sad reality is that if we simply say "The person with the most comments in a week gets a gold star", some joker will decide to make all of there comments one word at time, just so they can have their shinny gold star.
The first incarnation of this new "Hall of Fame" will be our best attempt at answering these questions, but like all things, it will be a work in progress, and with help from the community, if it is not the "best" on day one we'll get it there in time.
My First Journal Entry. The title seems fitting, the day after the world lost Dennis Ritchie.
First, Let me introduce myself. I am Dave Mueller, and I am currently the Lead Software Developer for Slashdot. Though I've been a long time reader of slashdot, I first came to work here in March of 2011. I've been writing PERL for a living since 1999, building web based applications since 2001, and working entirely in a Linux/Apache environment since 2003. I consider myself a real geek, I love SciFi, computers, electronics, robotics, and all things high tech. I am an amateur maker in my free time, and have started recording my personal projects on my web-site http://perljedi.com.
In my short time here, I have inherited the some times daunting task of not only maintianing slashdot, but also of helping to keep us (or put us back) on the leading edge of technology, news, and geek culture. With the parting of our illustrious founder CmdrTaco, we look to our readers, old and new to help us find our way into the future. I will be using my journal here to share insight into our team, and the processes we are going through everyday to find how best to server our readers, and the world.
My current project is a re-examination of our Hall of Fame. I hope by highlighting our best X at a finer level of detail (where X could be anything from "Most commented on Story", to "Most active moderator", or anything else within reason) we will be able to encourage people to engage more with the site. We want slashdot to be a place that is welcoming to geek's of all kinds, and I hope to show our readers that whether you have been reading slashdot since the 1990's, or since July, you too can help us understand and shape the community, and through it, the world around us. I hope that doesn't sound to wishy-washey or corny. But I am a true believer in what we are doing here, and I am quite passionate about geeks, geek culture, and our role in shaping the world.
I'm not sure how many people will read this, but I want all slashdotter's, old and new, to engage with my team and the editorial team to help us keep slashdot great.
Real programs don't eat cache.