Search Results

Search found 1745 results on 70 pages for 'probability theory'.

Page 39/70 | < Previous Page | 35 36 37 38 39 40 41 42 43 44 45 46  | Next Page >

  • Why job postings always looking for "rockstars?" [closed]

    - by Xepoch
    I have noticed a recent trend in requesting programmers who are rockstars. I get it, they're looking for someone who is really good at what they do. But why (pray) make the reference to a rockstar? Do these companies really want these traits as a real rockstar? Party all night and wake up to take care of quick business in the morning? Substance abuse, Narcissism with celebrity, Compensation well exceeding their management, Excellent at putting on a short-lived show, Entertainment instead of value, 1 hit (project) wonders or single-genre performers, Et cetera What is wrong with Senior or Principal Software Engineer who has an established and proven passion for the business? Rather do we mean quite the opposite, someone who: rolls up the sleeves and gets to work, takes appropriate direction and helps influence teams, programs in lessons' learned and proper practices, provides timely communication to the whole team, can code and understand multiple languages, understands the science and theory behind computation, Is there a trend to diversify the software engineering ranks? How many software rockstars can you hire before your band starts breaking up? Sure, there are lots of folks doing this stuff on their own, maybe even a rare few who do coding for show, but I wager the majority is for business. I don't see ads for rockstar accountants, or rockstar machinists, or rockstart CFOs. What makes the software programmer and their hiring departments lean towards this kind of job title?

    Read the article

  • Information about rendering, batches, the graphical card, performance etc. + XNA?

    - by Aidiakapi
    I know the title is a bit vague but it's hard to describe what I'm really looking for, but here goes. When it comes to CPU rendering, performance is mostly easy to estimate and straightforward, but when it comes to the GPU due to my lack of technical background information, I'm clueless. I'm using XNA so it'd be nice if theory could be related to that. So what I actually wanna know is, what happens when and where (CPU/GPU) when you do specific draw actions? What is a batch? What influence do effects, projections etc have? Is data persisted on the graphics card or is it transferred over every step? When there's talk about bandwidth, are you talking about a graphics card internal bandwidth, or the pipeline from CPU to GPU? Note: I'm not actually looking for information on how the drawing process happens, that's the GPU's business, I'm interested on all the overhead that precedes that. I'd like to understand what's going on when I do action X, to adapt my architectures and practices to that. Any articles (possibly with code examples), information, links, tutorials that give more insight in how to write better games are very much appreciated. Thanks :)

    Read the article

  • How much a programmer should read in order to keep himself updated? [closed]

    - by anything
    There are lots of technical books available. Below are few links which lists some good books If you could only have one programming related book on your bookshelf what would it be and why? What non-programming books should a programmer read to help develop programming/thinking skills? Best books on the theory and practice of software architecture? http://stackoverflow.com/questions/1711/what-is-the-single-most-influential-book-every-programmer-should-read ... and the list can go on and on and on. It will be really difficult to read all of the above mentioned books. I am not sure if its even possible for anyone to do that. Even if you filter it based on one's area of interest or work, list is still very large. .. and the technology keeps on changing (even more books :-( ) So, my question is how much a programmer should read lets say per year? How much hours one should put in such activities to keep oneself up to date? How do we find out the time required? PS: Average programmer reads less than one book per year (Code complete). What about the good programmers?

    Read the article

  • Shoring up deficiencies in a "home grown" programmer?

    - by JohnP
    I started out by teaching myself BASIC on a Vic 20, and in college (mid 80's) I had Fortran, Pascal, limited C, machine and assembler (With a smattering of COBOL). I didn't touch programming from approx 1989 to 1999. At that point, I was lucky enough to get hired as a Clipper programmer. Took me about 6 months to learn most of it, and by now (13 yrs) I'm pretty expert in it. I have also picked up Cold Fusion, some C#, some ASP, SQL, etc. I know programming structures, but in most languages I'm missing the esoterics, and I know my code could be much tighter. The problem is that I've learned what I needed to, to get the job done. This results in a lot of gaps in practical knowledge. I am also missing out on a TON of theory. Things like SRP, Refactoring, etc are alien terms. (Although I grok the intent after a short read). In addition, I am in the position now of teaching junior programmers the company and our software, and I don't want to pass on the knowledge gaps. I know this is somewhat of a subjective question and may be closed, but how do you go back and pick up what you've missed?

    Read the article

  • Can anyone point me to some open source directX rendering engines or frameworks? [on hold]

    - by Jim
    I'm completely new to graphics API programmming, but not at all new to the theory and principle operation of game engines and rendering engines. That being said, I want to do some experiments of rendering very dense geometry scenes in a basic rendering engine or game engine. I don't need a lot of bells and whistles. What I need is enough control that I can implement my own scene graph algorithms and control the rendering pipeline very specifically. My ideal candidate engine would be either a rendering engine or game engine with a modular design that might be ready to go out of the box but would be simple enough in case I need to rip out some of the guts in the rendering management and implement my own. It's a tough call because I'm right at the level where it's almost better to go from scratch, but there's no sense in having to build every single basic thing such as heirarchical transforms, etc. I just want to work with rendering optimization to push dense geometry for maximum FPS. Does anyone have a suggestion for an engine or basic framework to use? I requested DirectX in my title because I figured it would likely be better supported and less likely for me to run into some obscure less-documented problem. But OpenGL might be acceptable if the recommended framework was definitely better than my other options. EDIT: I should add that I really want GPU tessellation support (part of adding to the density of geometry detail).

    Read the article

  • Cannot usermod -L in LightDM scripts

    - by user95723
    I'm running Xubuntu 12.04 and use the LightDM. I want to restrict access to the machine as a kind of parental control. This is how it should work I hook in a script that executes just before the greeter comes up. Within that script some awk processing will read an entry in a config file and will trigger a usermod -L or usermod -U depending on whether the user is allowed to login. While user is logged, a cron job will count down the entry in the config and forces a xfce4-session-logout if time is up. A cron job running on a server will upload the "credits" on a daily base. How is this idea? That's theory, now for the problems It appears for some unknown reason, the usermod command is not executed, neither as part of a display-setup-script nor within the greeter-setup-script. I wrote a small sandbox script usermod -L johndoe 2error.txt touch /etc/blabla 2error.txt The script is executing, cause the blabla file is existing. That means that the script must have been executed with root privileges. error.txt is empty but the usermod command has just no effect. Is this a bug or a feature. What's wrong? Best regards and thank you Oli

    Read the article

  • Will I be able to get programming interviews at good software companies with a non-CS degree?

    - by friend
    I'll be graduating in a year, but I'll have a degree in Economics. I'm pretty much done with my Economics coursework, and by the time next year comes around I will have devoted 1.5 years to learning CS. I will have almost finished the requirements to graduate with a degree in CS, but unfortunately my school requires a science series that would add another 6-9 months of study if I were to try and get the degree (not to mention a max unit cap). I have or will have taken: Objected Oriented Programming Discrete Math Data structures Calculus through multivariable (doubt this matters at all) Linear Algebra (same) Computer Organization Operating Systems Computational Statistics (many data mining projects in R) Parallel Programming Programming Languages Databases Algorithms Compilers Artificial Intelligence I've done well in the ones I've taken, and I hope to do well in the rest, but will that matter if I can't say to the HR people that I have a CS degree? I'd be happy to get an internship at first too, so should I just apply as if I'm an intern and not looking for fulltime, and then try and parlay that into something? Sidenote if you have time -- Is a computer networks or theory of computation class important? Would it be worth taking either of those in lieu of a class on my list? edit -- I know this isn't AskReddit or College Confidential; I know there will be some outrage at posting a question like this. I'm merely looking for insight into a situation that I've been struggling with, and I think this is the absolute best place to find an answer to this question. Thanks.

    Read the article

  • PC doesn't POST with a certain model of PSU

    - by Core Xii
    I have this PC with an Asus P5N32-E SLI motherboard and an Intel Q6600 CPU. I got a new RX-5300 PSU, but it didn't work. The motherboard power LED is on fine When I switch power on, the PC powers up briefly (0.5-1 sec) then shuts down From there on out when I switch power on, it stays on, all fans and components seem to be receiving power, but the motherboard won't POST. No video output, no PC speaker beeps, nothing If I turn the hard switch on the PSU off and then back on again, go to step 2; The PC turns on and then off immediately again, and on subsequent power-ups it stays on but won't POST I disconnected every component but the motherboard, CPU and PSU. Still nothing. I tried three other models of PSU on this PC, both of higher (600) and lower (<300) wattage than the 530 on the RX-5300 and they all work fine. At this point I was convinced the PSU was faulty so I returned it. When the replacement arrived, it behaved exactly the same. So two different units of RX-5300 both with the same symptoms, neither working with this motherboard + CPU. Yet, three other models of PSU work perfectly fine. The PC store couldn't reproduce my problem with the returned PSU. I tried resetting the CMOS with the jumper. I tried with both the 4 and 4+4 (with and without the extra +4 connected) CPU connectors (curiously the RX-5300 comes equipped with both). Could it be a statistical probability that I get two units of the same model of PSU that are faulty in the exact same manner? Could the RX-5300 model itself be somehow incompatible with this motherboard? I was under the impression that PSUs were pretty much universal so long as you have the wattage. Could the motherboard be broken in some such a way as to work with certain PSUs but not others? What's going on here?

    Read the article

  • Using Queries with Coherence Read-Through Caches

    - by jpurdy
    Applications that rely on partial caches of databases, and use read-through to maintain those caches, have some trade-offs if queries are required. Coherence does not support push-down queries, so queries will apply only to data that currently exists in the cache. This is technically consistent with "read committed" semantics, but the potential absence of data may make the results so unintuitive as to be useless for most use cases (depending on how much of the database is held in cache). Alternatively, the application itself may manually "push down" queries to the database, either retrieving results equivalent to querying the cache directly, or may query the database for a key set and read the values from the cache (relying on read-through to handle any missing values). Obviously, if the result set is too large, reading through the cache may cause significant thrashing. It's also worth pointing out that if the cache is asynchronously synchronized with the database (perhaps via database change listener), that an application may commit a transaction to the database, then generate a key set from the database via a query, then read cache entries through the cache, possibly resulting in a race condition where the application sees older data than it had previously committed. In theory this is not problematic but in practice it is very unintuitive. For this reason it often makes sense to invalidate the cache when updating the database, forcing the next read-through to update the cache.

    Read the article

  • How to disguise a serverside mob as another?

    - by Shaun Wild
    I've been working a Minecraft sever mod and i want to be able to add a new entity to the server, but then make the server send the packets to the client, imitating another mob, for example.. Lets say say i have EntityPlayerNPC.class, what i want to do is have all of the packets that get sent to the client look like they are from that of another player which is on the player, therefore allowing me to add custom NPC's... Thinking about the theory i'm sure this can be done. I've tried looking around for where the packets are being sent from and whatnot, can anyone think up a solution? edit: i tried adding a new constructor to the Packet20NamedEntitySpawn class like so: public Packet20NamedEntitySpawn(String username, EntityLiving e){ this.entityId = 0; this.name = username; this.xPosition = MathHelper.floor_double(e.posX * 32.0D); this.yPosition = MathHelper.floor_double(e.posY * 32.0D); this.zPosition = MathHelper.floor_double(e.posZ * 32.0D); this.rotation = (byte)((int)(e.rotationYaw * 256.0F / 360.0F)); this.pitch = (byte)((int)(e.rotationPitch * 256.0F / 360.0F)); this.metadata = e.getDataWatcher(); } unfortunatley, that didn't work :(

    Read the article

  • How to better explain complex software process in software specs?

    - by Lostsoul
    I'm really struggling with my software specs. I am not a professional programmer but enjoy doing it for fun and made some software that I want to sell later but I'm not happy with the code quality. So I wanted to hire a real developer to rewrite my software in a more professional way so it will be maintainable by other developers in the future. I read and found some sample specs and made my own by applying their structure to my document and wanted to get my developer friend to read it and give me advice. After an hour and a half he understood exactly what I was trying to do and how I did it(my algorithms,stack,etc.). How can I get better at explaining things to developers? I add many details and explanations for everything(including working code) but I'm unsure the best way I can learn to pass detailed domain knowledge(my software applies big data, machine learning, graph theory to finance). My end goal is to get them to understand as much as possible from the document and then ask anything they do not understand, but right now it seems they need to extract alot of information from me. How can I get better at communicating domain knowledge to developers?

    Read the article

  • Can my employer force me to backup my personal machine? [closed]

    - by Eric B
    Here's the background: Approximately 1.25 years ago, the company I work for was acquired by a larger 400 person company. Before acquisition (and today still) we are all remote employees using our own personal hardware for work-related duties (coding, email, etc). We are approximately 15 employees within the larger organization. Some time after acquisition, the now owning company was slapped with a civil lawsuit. Part of this lawsuit (discovery) is requiring them to retrieve & store from us any related information. Because we were a separate company up until acquisition, there is a high probability that our personal machines might contain information about what the lawsuit alleges (email, documents, chat logs?, etc). Obviously, this depends largely on the person's job function (engineer vs. customer support vs. CEO). All employees are being required to comply. Since acquisition (1.25 yrs), the new company has not provided us with company laptops/desktops. We continue to use personal hardware, licenses, etc for work. Email is via POP3s and not hanging around on the mail server - it's on everyone's client. Documents are spread across personal machines. So, now they want us each to backup our complete personal machines. They are allowing us to create a "personal" folder where we can place personal documents. That single folder will be excluded from backup. Of course, that means total re-arrangement of documents, etc. For most of us, 99% of the data on the machine is NOT related to work. So, what's the consensus? Should we comply? What is their recourse if we do not?

    Read the article

  • Where can someone store >100GB of pictures online? [closed]

    - by sbi
    A person who is not very computer-savvy needs to store 130GB of photos. The key parameters are: an non-negligible probability that the company selling the storage will be existing, and the data accessible, for at least five years data should be considered safe once uploaded reasonable terms of service: google drive reserving the right to literally do anything they want with their user's data is not acceptable; the possibility that the CIA might look at those pictures is not considered a threat easy to use from Windows, preferably as a drive no nerve-wracking limitations ("cannot upload 10GB/day" or "files 500MB" etc.) that serve no purpose other than pushing the user to the next-higher price plan some upgrade plan: there's currently 10-30GB of new photos per year, with a tendency to increase, which might bust a 150GB limit next January ability to somehow sort the pictures: currently they are sorted into folders, but something alike (tags) would be just as good, if easy enough to apply of course, the pricing is important (although there's a reason this is the last bullet; reasonable data safety is considered more important) Nice to have, but not necessary features would be: additional features related to photos (thumbnail generation, album sharing etc.) access from web and other platforms than Windows (smart phones) Let me stress this again: The person in need of that is able to copy pictures from the camera to the computer, can copy files in the explorer, and uses a web email service. That's about it, there's almost no understanding of what happens under the hood.

    Read the article

  • Provide A Scrolling "Camera" View Over A 2D Game Map

    - by BitCrash
    I'm in the process of attempting to create a 2D MMO type game with Kryonet and some basic sprites, mostly for my own learning. I have the back end set up great (By my standards) and I'm moving on to actually getting some things drawn onto the map. I cannot for the life of me figure out a solid way to have a "Camera" follow a player around a large area. The view pane for the game is 640 x 480 pixels, and each tile is 32x32 pixels (Thats 20 tiles wide and 15 high for the viewpane) I have tried a couple things to do this, but they did not seem to work out so well. I had a JScrollPane with 9 "Viewpane"-sized canvases in it, and tried to have the JScrollPane move in accordance with the player. The issue came when I reached the end of the JScrollPane. I tried to "Flip" canvases, sending the canvas currrently drawing the player to the middle of the 9 and load the corresponding maps onto the other ones. It was slow and worked poorly. I'm looking for any advice or previous experience with this; any ideas? Thank you! Edit and Clarification: I did not mean to mention Kryonet, I was merely providing peripheral information in case there was something that would help which I could not foresee. Instead of having an array of 9 canvases, why not just have one large canvas loading a large map every once in a while? I'm willing to have "load times" where as with the canvas array I would have none (in theory) to give the user a smooth experience. I could just change the size and location of the map with a modified setBounds() call on the canvas in a layered pane (layered because I have hidden swing items, like inventories and stuff) I'll try it out and post here how it goes for people asking the same question.

    Read the article

  • Implementing Explosions

    - by Xkynar
    I want to add explosions to my 2D game, but im having a hard time with the architecture. Several game elements might be responsible for explosions, like, lets say, explosive barrels and bullets (and there might be chain reactions with close barrels). The only options i can come up with are: 1 - Having an array of explosions and treat them as a game element as important as any other Pros: Having a single array which is updated and drawn with all the other game element arrays makes it more organized and simple to update, and the explosive barrels at a first glance would be easy to create, simply by passing the explosion array as a pointer to each explosive barrel constructor Cons: It might be hard for the bullets to add an explosion to the vector, since bullets are shot by a Weapon class which is located in every mob, so lets say, if i create a new enemy and add it to the enemy array, that enemy will have a weapon and functions to be able to use it, and if i want the weapon (rocket launcher in this case) to have access to the explosions array to be able to add a new one, id have to pass the explosion array as a pointer to the enemy, which would then pass it to the weapon, which would pass it to the bullets (ugly chain). Another problem I can think of is a little more weird: If im checking the collisions between explosions and barrels (so i create a chain reaction) and i detect an explosion colliding with a barrel, if i add a new explosion while im iterating the explosions java will trow an exception. So this is kinda annoying, i cant iterate through the explosions and add a new explosion, i must do it in another way... The other way which isnt really well thought yet is to just add an explosive component to every element that might explode so that when it dies, it explodes or something, but i dont have good ways on implementing this theory either Honestly i dont like either the solutions so id like to know how is it usually done by actual game developers, sorry if my problem seems trivial and dumb.

    Read the article

  • Which is the best way to catch an expiring domain name? [closed]

    - by newspeak
    I know a similar question has been asked, but I really don't know what to do. There is this .com domain which is currently on redemption period and should likely be available again within a month. I was wondering which is the best way to get it at a reasonable price. I don't think it's a highly valuable domain, it shows to have very bad ranking and has 0 exact same searches according to adwords. Why it is valuable to me is very simple: I have a project responding to this name. I already own the .net domain and would love to have the .com. I discovered the domain was going to be available thanks to an email I received by a backorder site. I did some research and these guys have a bad reputation on the web. I did further research and found that more reputable (at least in theory) companies should be the likes of snapnames, pool, namejet, godaddy, etc. I am a bit suspicious using these drop cathing services: What if they shill bids? What if they make it go into auction even if I'm the only person interested? What if I raise attention and interest to the domain by backordering? I just would rather wait for it to be deleted and available again to register it manually. It is really not an interesting domain name, and I don't think anyone would care to have it. But what if the domain is already being watched by the domain industry sharks? I did a whois research and my desired domain nameserves point to domcollect.com, which appears to be an auction site. What if I decide to wait for manual registration and I miss the chance to get it? I'm willing to spend the 60/70$ fees these sites require, but not really more than that. Suggestions? Thank you very much. I'm a bit confused and undecided.

    Read the article

  • How can I get better at explaining complex software processes to developers?

    - by Lostsoul
    I'm really struggling with my software specs. I am not a professional programmer but enjoy doing it for fun and made some software that I want to sell later but I'm not happy with the code quality. So I wanted to hire a real developer to rewrite my software in a more professional way so it will be maintainable by other developers in the future. I read and found some sample specs and made my own by applying their structure to my document and wanted to get my developer friend to read it and give me advice. After an hour and a half he understood exactly what I was trying to do and how I did it(my algorithms,stack,etc.). How can I get better at explaining things to developers? I add many details and explanations for everything(including working code) but I'm unsure the best way I can learn to pass detailed domain knowledge(my software applies big data, machine learning, graph theory to finance). My end goal is to get them to understand as much as possible from the document and then ask anything they do not understand, but right now it seems they need to extract alot of information from me. How can I get better at communicating domain knowledge to developers?

    Read the article

  • Iptables working strangely

    - by user109985
    I have Ubuntu 12.04 x64 installed in my laptop. I'm quite new to linux, and I wanted to specify certain rules for the iptables firewall. This is the saved config of my iptables: *mangle :PREROUTING ACCEPT [0:0] :INPUT ACCEPT [0:0] :FORWARD ACCEPT [0:0] :OUTPUT ACCEPT [0:0] :POSTROUTING ACCEPT [0:0] COMMIT *nat :PREROUTING ACCEPT [0:0] :INPUT ACCEPT [0:0] :OUTPUT ACCEPT [0:0] :POSTROUTING ACCEPT [0:0] -A PREROUTING -p tcp -m tcp --dport 3306 -j DNAT --to-destination 192.168.1.100:3306 -A PREROUTING -p tcp -m tcp --dport 11002 -j DNAT --to-destination 192.168.1.100:11002 -A PREROUTING -p tcp -m tcp --dport 13000 -j DNAT --to-destination 192.168.1.100:13000 -A PREROUTING -p tcp -m tcp --dport 13001 -j DNAT --to-destination 192.168.1.100:13001 -A PREROUTING -p tcp -m tcp --dport 13002 -j DNAT --to-destination 192.168.1.100:13002 -A PREROUTING -p tcp -m tcp --dport 13003 -j DNAT --to-destination 192.168.1.100:13003 -A PREROUTING -p tcp -m tcp --dport 13004 -j DNAT --to-destination 192.168.1.100:13004 -A PREROUTING -p tcp -m tcp --dport 13061 -j DNAT --to-destination 192.168.1.100:13061 -A PREROUTING -p tcp -m tcp --dport 13099 -j DNAT --to-destination 192.168.1.100:13099 -A POSTROUTING -j MASQUERADE COMMIT *filter :INPUT DROP [0:0] :FORWARD ACCEPT [0:0] :OUTPUT ACCEPT [0:0] -A INPUT -i eth0 -p tcp -m tcp --dport 3306 -j ACCEPT -A INPUT -i eth0 -p tcp -m tcp --dport 80 -j ACCEPT COMMIT In theory, what it does is to drop all connections in all ports except for http, mysql and few other ports. Moreover, it redirects all mysql and 13000-13004..etc port connections to a specific local ip in the same port, which is 192.168.1.100. But what I find strange is that when my firewall is active, it blocks absolutely all the input connections, even those which must not block (http, mysql). In fact, I'm literally blocked and I can't establish any external connection. What am I doing wrong? PS: I tested the firewall without those redirections, and it still block all inputs, so I suppose it's not the problem.

    Read the article

  • How to learn to program [on hold]

    - by user94914
    I went to a community college and got a degree in computer science, but I found out I only learn very little about programming. As a result I landed byself a office assistant work (for a year now), I want to study on my own and apply for some internship / very entry level development job. I am wondering how should a person learn to program now? I feel that I might not doing it correctly, I understand everyone has a different approach, but I am really clueless on what to do, as it seems I am 5-10 years away. 1) Read the old college programming textbook cover to cover, learn every single concepts, do all the practice problems and master them (1-2 times until error free). Currently reading this java book 2) Work on any project, keep on googling and reading tutorials (including the books on that specific language). I have been doing 1, but the progress is really slow, about 2-5 pages / hour, over a 1000+ page book, I felt really discouraged. I have a few of them to go through (data struction, analyis algorthim, computer theory, operating system.) I wonder is this the right method to do? I know it is going to take time, but I am hoping to get some advice from current programmers.

    Read the article

  • Another Marketing Conference, part one – the best morning sessions.

    - by Roger Hart
    Yesterday I went to Another Marketing Conference. I honestly can’t tell if the title is just tipping over into smug, but in the balance of things that doesn’t matter, because it was a good conference. There was an enjoyable blend of theoretical and practical, and enough inter-disciplinary spread to keep my inner dilettante grinning from ear to ear. Sure, there was a bumpy bit in the middle, with two back-to-back sales pitches and a rather thin overview of the state of the web. But the signal:noise ratio at AMC2012 was impressively high. Here’s the first part of my write-up of the sessions. It’s a bit of a mammoth. It’s also a bit of a mash-up of what was said and what I thought about it. I’ll add links to the videos and slides from the sessions as they become available. Although it was in the morning session, I’ve not included Vanessa Northam’s session on the power of internal comms to build brand ambassadors. It’ll be in the next roundup, as this is already pushing 2.5k words. First, the important stuff. I was keeping a tally, and nobody said “synergy” or “leverage”. I did, however, hear the term “marketeers” six times. Shame on you – you know who you are. 1 – Branding in a post-digital world, Graham Hales This initially looked like being a sales presentation for Interbrand, but Graham pulled it out of the bag a few minutes in. He introduced a model for brand management that was essentially Plan >> Do >> Check >> Act, with Do and Check rolled up together, and went on to stress that this looks like on overall business management model for a reason. Brand has to be part of your overall business strategy and metrics if you’re going to care about it at all. This was the first iteration of what proved to be one of the event’s emergent themes: do it throughout the stack or don’t bother. Graham went on to remind us that brands, in so far as they are owned at all, are owned by and co-created with our customers. Advertising can offer a message to customers, but they provide the expression of a brand. This was a preface to talking about an increasingly chaotic marketplace, with increasingly hard-to-manage purchase processes. Services like Amazon reviews and TripAdvisor (four presenters would make this point) saturate customers with information, and give them a kind of vigilante power to comment on and define brands. Consequentially, they experience a number of “moments of deflection” in our sales funnels. Our control is lessened, and failure to engage can negatively-impact buying decisions increasingly poorly. The clearest example given was the failure of NatWest’s “caring bank” campaign, where staff in branches, customer support, and online presences didn’t align. A discontinuity of experience basically made the campaign worthless, and disgruntled customers talked about it loudly on social media. This in turn presented an opportunity to engage and show caring, but that wasn’t taken. What I took away was that brand (co)creation is ongoing and needs monitoring and metrics. But reciprocally, given you get what you measure, strategy and metrics must include brand if any kind of branding is to work at all. Campaigns and messages must permeate product and service design. What that doesn’t mean (and Graham didn’t say it did) is putting Marketing at the top of the pyramid, and having them bawl demands at Product Management, Support, and Development like an entitled toddler. It’s going to have to be collaborative, and session 6 on internal comms handled this really well. The main thing missing here was substantiating data, and the main question I found myself chewing on was: if we’re building brands collaboratively and in the open, what about the cultural politics of trolling? 2 – Challenging our core beliefs about human behaviour, Mark Earls This was definitely the best show of the day. It was also some of the best content. Mark talked us through nudging, behavioural economics, and some key misconceptions around decision making. Basically, people aren’t rational, they’re petty, reactive, emotional sacks of meat, and they’ll go where they’re led. Comforting stuff. Examples given were the spread of the London Riots and the “discovery” of the mountains of Kong, and the popularity of Susan Boyle, which, in turn made me think about Per Mollerup’s concept of “social wayshowing”. Mark boiled his thoughts down into four key points which I completely failed to write down word for word: People do, then think – Changing minds to change behaviour doesn’t work. Post-rationalization rules the day. See also: mere exposure effects. Spock < Kirk - Emotional/intuitive comes first, then we rationalize impulses. The non-thinking, emotive, reactive processes run much faster than the deliberative ones. People are not really rational decision makers, so  intervening with information may not be appropriate. Maximisers or satisficers? – Related to the last point. People do not consistently, rationally, maximise. When faced with an abundance of choice, they prefer to satisfice than evaluate, and will often follow social leads rather than think. Things tend to converge – Behaviour trends to a consensus normal. When faced with choices people overwhelmingly just do what they see others doing. Humans are extraordinarily good at mirroring behaviours and receiving influence. People “outsource the cognitive load” of choices to the crowd. Mark’s headline quote was probably “the real influence happens at the table next to you”. Reference examples, word of mouth, and social influence are tremendously important, and so talking about product experiences may be more important than talking about products. This reminded me of Kathy Sierra’s “creating bad-ass users” concept of designing to make people more awesome rather than products they like. If we can expose user-awesome, and make sharing easy, we can normalise the behaviours we want. If we normalize the behaviours we want, people should make and post-rationalize the buying decisions we want.  Where we need to be: “A bigger boy made me do it” Where we are: “a wizard did it and ran away” However, it’s worth bearing in mind that some purchasing decisions are personal and informed rather than social and reactive. There’s a quadrant diagram, in fact. What was really interesting, though, towards the end of the talk, was some advice for working out how social your products might be. The standard technology adoption lifecycle graph is essentially about social product diffusion. So this idea isn’t really new. Geoffrey Moore’s “chasm” idea may not strictly apply. However, his concepts of beachheads and reference segments are exactly what is required to normalize and thus enable purchase decisions (behaviour change). The final thing is that in only very few categories does a better product actually affect purchase decision. Where the choice is personal and informed, this is true. But where it’s personal and impulsive, or in any way social, “better” is trumped by popularity, endorsement, or “point of sale salience”. UX, UCD, and e-commerce know this to be true. A better (and easier) experience will always beat “more features”. Easy to use, and easy to observe being used will beat “what the user says they want”. This made me think about the astounding stickiness of rational fallacies, “common sense” and the pathological willful simplifications of the media. Rational fallacies seem like they’re basically the heuristics we use for post-rationalization. If I were profoundly grimy and cynical, I’d suggest deploying a boat-load in our messaging, to see if they’re really as sticky and appealing as they look. 4 – Changing behaviour through communication, Stephen Donajgrodzki This was a fantastic follow up to Mark’s session. Stephen basically talked us through some tactics used in public information/health comms that implement the kind of behavioural theory Mark introduced. The session was largely about how to get people to do (good) things they’re predisposed not to do, and how communication can (and can’t) make positive interventions. A couple of things stood out, in particular “implementation intentions” and how they can be linked to goals. For example, in order to get people to check and test their smoke alarms (a goal intention, rarely actualized  an information campaign will attempt to link this activity to the clocks going back or forward (a strong implementation intention, well-actualized). The talk reinforced the idea that making behaviour changes easy and visible normalizes them and makes them more likely to succeed. To do this, they have to be embodied throughout a product and service cycle. Experiential disconnects undermine the normalization. So campaigns, products, and customer interactions must be aligned. This is underscored by the second section of the presentation, which talked about interventions and pre-conditions for change. Taking the examples of drug addiction and stopping smoking, Stephen showed us a framework for attempting (and succeeding or failing in) behaviour change. He noted that when the change is something people fundamentally want to do, and that is easy, this gets a to simpler. Coordinated, easily-observed environmental pressures create preconditions for change and build motivation. (price, pub smoking ban, ad campaigns, friend quitting, declining social acceptability) A triggering even leads to a change attempt. (getting a cold and panicking about how bad the cough is) Interventions can be made to enable an attempt (NHS services, public information, nicotine patches) If it succeeds – yay. If it fails, there’s strong negative enforcement. Triggering events seem largely personal, but messaging can intervene in the creation of preconditions and in supporting decisions. Stephen talked more about systems of thinking and “bounded rationality”. The idea being that to enable change you need to break through “automatic” thinking into “reflective” thinking. Disruption and emotion are great tools for this, but that is only the start of the process. It occurs to me that a great deal of market research is focused on determining triggers rather than analysing necessary preconditions. Although they are presumably related. The final section talked about setting goals. Marketing goals are often seen as deriving directly from business goals. However, marketing may be unable to deliver on these directly where decision and behaviour-change processes are involved. In those cases, marketing and communication goals should be to create preconditions. They should also consider priming and norms. Content marketing and brand awareness are good first steps here, as brands can be heuristics in decision making for choice-saturated consumers, or those seeking education. 5 – The power of engaged communities and how to build them, Harriet Minter (the Guardian) The meat of this was that you need to let communities define and establish themselves, and be quick to react to their needs. Harriet had been in charge of building the Guardian’s community sites, and learned a lot about how they come together, stabilize  grow, and react. Crucially, they can’t be about sales or push messaging. A community is not just an audience. It’s essential to start with what this particular segment or tribe are interested in, then what they want to hear. Eventually you can consider – in light of this – what they might want to buy, but you can’t start with the product. A community won’t cohere around one you’re pushing. Her tips for community building were (again, sorry, not verbatim): Set goals Have some targets. Community building sounds vague and fluffy, but you can have (and adjust) concrete goals. Think like a start-up This is the “lean” stuff. Try things, fail quickly, respond. Don’t restrict platforms Let the audience choose them, and be aware of their differences. For example, LinkedIn is very different to Twitter. Track your stats Related to the first point. Keeping an eye on the numbers lets you respond. They should be qualified, however. If you want a community of enterprise decision makers, headcount alone may be a bad metric – have you got CIOs, or just people who want to get jobs by mingling with CIOs? Build brand advocates Do things to involve people and make them awesome, and they’ll cheer-lead for you. The last part really got my attention. Little bits of drive-by kindness go a long way. But more than that, genuinely helping people turns them into powerful advocates. Harriet gave an example of the Guardian engaging with an aspiring journalist on its Q&A forums. Through a series of serendipitous encounters he became a BBC producer, and now enthusiastically speaks up for the Guardian community sites. Cultivating many small, authentic, influential voices may have a better pay-off than schmoozing the big guys. This could be particularly important in the context of Mark and Stephen’s models of social, endorsement-led, and example-led decision making. There’s a lot here I haven’t covered, and it may be worth some follow-up on community building. Thoughts I was quite sceptical of nudge theory and behavioural economics. First off it sounds too good to be true, and second it sounds too sinister to permit. But I haven’t done the background reading. So I’m going to, and if it seems to hold real water, and if it’s possible to do it ethically (Stephen’s presentations suggests it may be) then it’s probably worth exploring. The message seemed to be: change what people do, and they’ll work out why afterwards. Moreover, the people around them will do it too. Make the things you want them to do extraordinarily easy and very, very visible. Normalize and support the decisions you want them to make, and they’ll make them. In practice this means not talking about the thing, but showing the user-awesome. Glib? Perhaps. But it feels worth considering. Also, if I ever run a marketing conference, I’m going to ban speakers from using examples from Apple. Quite apart from not being consistently generalizable, it’s becoming an irritating cliché.

    Read the article

  • Quick Quips on QR Codes

    - by Tim Dexter
    Yes, I'm an alliterating all-star; I missed my calling as a newspaper headline writer. I have recently received questions from several folks on support for QR codes. You know them they are everywhere you look, even here! How does Publisher handle QR codes then? In theory, exactly the same way we handle any other 2D barcode font. We need the font file, a mapping entry and an encoding class. With those three pieces we can embed QR codes into any output. To test the theory, I went off to IDAutomation, I have worked with them and many customers over the years and their fonts and encoders have worked great and have been very reliable. They kindly provide demo fonts which has made my life so much easier to be able to write posts like this. Their QR font and encoder is a little tough to find. I started here and then hit the Demo Now button. On the next page I hit the right hand Demo Now button. In the resulting zip file you'll need two files: AdditionalFonts.zip >> Automation2DFonts >> TrueType >> IDAutomation2D.ttf Java Class Encoder >> IDAutomation_JavaFontEncoder_QRCode.jar - the QRBarcodeExample.java is useful to see how to call the encoder. The font file needs to be installed into the windows/fonts directory, just copy and paste it in using file explorer and windows will install it for you. Remember, we are using the demo font here and you'll see if you get your phones decoder to looks a the font above there is a fixed string 'DEMO' at the beginning. You want that removed? Go buy the font from the IDAutomation folks. The Encoder Next you need to create your encoding wrapper class. Publisher does ship a class but its compiled and I do not recommend trying to modify it, you can just build your own. I have loaded up my class here. You do not need to be a java guru, its pretty straightforward. I'd recommend a java IDE like JDeveloper from a convenience point of view. I have annotated my class and added a main method to it so you can test your encoders from JDeveloper without having to deploy them first. You can load up the project form the zip file straight into JDeveloper.Next, take a look at IDAutomation's example java class and you'll see: QRCodeEncoder qre=new QRCodeEncoder();  String DataToEncode = "IDAutmation Inc.";  boolean ApplyTilde = false;  int EncodingMode = 0;  int Version = 0;  int ErrorCorrectionLevel = 0;  System.out.println( qre.FontEncode(DataToEncode, ApplyTilde, EncodingMode, Version, ErrorCorrectionLevel) ); You'll need to check what settings you need to set for the ApplyTilde, EncodingMode, Version and ErrorCorrectionLevel. They are covered in the user guide from IDAutomation here. If you do not want to hard code the values in the encoder then you can quite easily externalize them and read the values from a text file. I have not covered that scenario here, I'm going with IDAutomation's defaults and my phone app is reading the fonts no problem. Now you know how to call the encoder, you need to incorporate it into your encoder wrapper class. From my sample class:       Class[] clazz = new Class[] { "".getClass() };        ENCODERS.put("code128a",mUtility.getClass().getMethod("code128a", clazz));       ENCODERS.put("code128b",mUtility.getClass().getMethod("code128b", clazz));       ENCODERS.put("code128c",mUtility.getClass().getMethod("code128c", clazz));       ENCODERS.put("qrcode",mUtility.getClass().getMethod("qrcode", clazz)); I just added a new entry to register the encoder method 'qrcode' (in red). Then I created a new method inside the class to call the IDAutomation encoder. /** Call to IDAutomations QR Code encoder. Passing the data to encode      Returning the encoded string to the template for formatting **/ public static final String qrcode (String DataToEncode) {   QRCodeEncoder qre=new QRCodeEncoder();    boolean ApplyTilde = false;    int EncodingMode = 0;    int Version = 0;    int ErrorCorrectionLevel = 0; return qre.FontEncode(DataToEncode, ApplyTilde, EncodingMode, Version, ErrorCorrectionLevel); } Almost the exact same code in their sample class. The DataToEncode string is passed in rather than hardcoded of course. With the class done you can now compile it, but you need to ensure that the IDAutomation_JavaFontEncoder_QRCode.jar is in the classpath. In JDeveloper, open the project properties >> Libraries and Classpaths and then add the jar to the list. You'll need the publisher jars too. You can find those in the jlib directory in your Template Builder for Word directory.Note! In my class, I have used package oracle.psbi.barcode; As my package spec, yours will be different but you need to note it for later. Once you have it compiling without errors you will need to generate a jar file to keep it in. In JDeveloper highlight your project node >> New >> Deployment Profile >> JAR file. Once you have created the descriptor, just take the defaults. It will tell you where the jar is located. Go get it and then its time to copy it and the IDAutomation jar into the Template Builder for Word directory structure. Deploying the jars On your windows machine locate the jlib directory under the Template Builder for Word install directory. On my machine its here, F:\Program Files\Oracle\BI Publisher\BI Publisher Desktop\Template Builder for Word\jlib. Copy both of the jar files into the directory. The next step is to get the jars into the classpath for the Word plugin so that Publisher can find your wrapper class and it can then find the IDAutomation encoder. The most consistent way I have found so far, is to open up the RTF2PDF.jar in the same directory and make some mods. First make a backup of the jar file then open it using winzip or 7zip or similar and get into the META-INF directory. In there is a file, MANIFEST.MF. This contains the classpath for the plugin, open it in an editor and add the jars to the end of the classpath list. In mine I have: Manifest-Version: 1.0 Class-Path: ./activation.jar ./mail.jar ./xdochartstyles.jar ./bicmn.jar ./jewt4.jar ./share.jar ./bipres.jar ./xdoparser.jar ./xdocore.jar ./xmlparserv2.jar ./xmlparserv2-904.jar  ./i18nAPI_v3.jar ./versioninfo.jar ./barcodejar.jar ./IDAutomation_JavaFontEncoder_QRCode.jar Main-Class: RTF2PDF I have put in carriage returns above to make the Class-Path: entry more readable, make sure yours is all on one line. Be sure to use the ./ as a prefix to the jar name. Ensure the file is saved inside the jar file 7zip and winzip both have popups asking if you want to update the file in the jar file.Now you have the jars on the classpath, the Publisher plugin will be able to find our classes at run time. Referencing the Font The next step is to reference the font location so that the rendering engine can find it and embed a subset into the PDF output. Remember the other output formats rely on the font being present on the machine that is opening the document. The PDF is the only truly portable format. Inside the config directory under the Template Builder for Word install directory, mine is here, F:\Program Files\Oracle\BI Publisher\BI Publisher Desktop\Template Builder for Word\config. You'll find the file, 'xdo example.cfg'. Rename it to xdo.cfg and open it in a text editor. In the fonts section, create a new entry:       <font family="IDAutomation2D" style="normal" weight="normal">              <truetype path="C:\windows\fonts\IDAutomation2D.ttf" />       </font> Note, 'IDAutomation2D' (in red) is the same name as you can see when you open MSWord and look for the QRCode font. This must match exactly. When Publisher looks at the fonts in the RTF template at runtime it will see 'IDAutomation2D' it will then look at its font mapping entries to find where that font file resides on the disk. If the names do not match or the font is not present then the font will not get used and it will fall back on Helvetica. Building the Template Now you have the data encoder and the font in place and mapped; you can use it in the template. The two commands you will need to have present are: <?register-barcode-vendor:'ENCODER WRAPPER CLASS'; 'ENCODER NAME'?> for my encoder I have: <?register-barcode-vendor:'oracle.psbi.barcode.BarcodeUtil'; 'MyBarcodeEncoder'?> Notice the two parameters for the command. The first provides the package 'path' and class name (remember I said you need to remember that above.)The second is the name of the encoder, in my case 'MyBarcodeEncoder'. Check my full encoder class in the zip linked below to see where I named it. You can change it to something else, no problem.This command needs to be near the top of the template. The second command is the encoding command: <?format-barcode:DATAT_TO_ENCODE;'ENCODER_METHOD_NAME';'ENCODER_NAME'?> for my command I have <?format-barcode:DATATEXT;'qrcode';'MyBarcodeEncoder'?>DATATEXT is the XML element that contains the text to be encoded. If you want to hard code a piece of text just surround it with single quotes. qrcode is the name of my encoder method that calls the IDAutomation encoder. Remember this.MyBarcodeEncoder is the name of my encoder. Repetition? Yes but its needed again. Both of these commands are put inside their own form fields. Do not apply the QRCode font to the second field just yet. Lets make sure the encoder is working. Run you template with some data and you should get something like this for your encoded data: AHEEEHAPPJOPMOFADIPFJKDCLPAHEEEHA BNFFFNBPJGMDIDJPFOJGIGBLMPBNFFFNB APIBOHFJCFBNKHGGBMPFJFJLJBKGOMNII OANKPJFFLEPLDNPCLMNGNIJIHFDNLJFEH FPLFLHFHFILKFBLOIGMDFCFLGJGOPJJME CPIACDFJPBGDODOJCHALJOBPECKMOEDDF MFFNFNEPKKKCHAIHCHPCFFLDAHFHAGLMK APBBBPAPLDKNKJKKGIPDLKGMGHDDEPHLN HHHHHHHPHPHHPHPPHPPPPHHPHHPHPHPHP Grooovy huh? If you do not get the encoded text then go back and check that your jars are in the right spot and that you have the MANIFEST.MF file updated correctly. Once you do get the encoded text, highlight the field and apply the IDAutomation2D font to it. Then re-run the report and you will hopefully see the QR code in your output. If not, go back and check the xdo.cfg entry and make sure its in the right place and the font location is correct. That's it, you now have QR codes in Publisher outputs. Everything I have written above, has been tested with the 5.6.3, 10.1.3.4.2 codelines. I'll be testing the 11g code in the next day or two and will update you with any changes. One thing I have not covered yet and will do in the next few days is how to deploy all of this to your server. Look out for a follow up post. One note on the apparent white lines in the font (see the image above). Once printed they disappear and even viewing the code on a screen with the white lines, my phone app is still able to read and interpret the contents no problem. I have zipped up my encoder wrapper class as a JDeveloper 11.1.1.6 project here. Just dig into the src directories to find the BarcodeUtil.java file if you just want the code. I have put comments into the file to hopefully help the novice java programmer out. Happy QR'ing!

    Read the article

  • What does WIX's CloseApplication functionality do and how would can the application respond to such

    - by tronda
    In the WIX setup I've got, when upgrading the application I have set a requirement to close down applications which might hold on to files which needs to be updated: <util:CloseApplication Id="CloseMyApp" Target="[MyAppExe]" CloseMessage="no" Description="!(loc.MyAppStillRunning)" RebootPrompt="no" ElevatedCloseMessage="no" /> The application on the other hand will capture closing down the window with a "user friendly" dialog box where the user can confirm that he or she wants to close down the application. I've experienced problems with the CloseApplication, and one theory is that the dialog box stops the application from closing. So the question is: Could this be a possible problem? If so - how can I have this confirmation dialog box and still behave properly when the installer asks the application to close down? Must I listen to Win32 messages (such as WM_QUIT/WM_CLOSE) or is there a .NET API which I can use to respond properly to these events?

    Read the article

  • Why won't this Apache modrewrite RewriteRule work?

    - by Jason Rhodes
    I'm trying to get Apache mod rewrite to work on my local machine. I'm running OSX with PHP 5 and the Apache mod rewrite module is enabled. I have a directory called localhost/~Jason/hfh/admin with various PHP includes called based on a $_GET variable. I want to let users type (in theory) localhost/~Jason/hfh/admin/pages and have that URL stay in the address bar, while what gets displayed is localhost/~Jason/hfh/admin/?admin=pages So. I've created a .htaccess file that sits in the /hfh directory. Inside, I've put this mod rewrite text: RewriteEngine On RewriteRule ^admin/([^/.]+)/?$ admin/?admin=$1 [L] When I go to the browser and type localhost/~Jason/hfh/admin/pages I get a "Problem loading page" error, and Firefox says, "Oops. Firefox can't load this page for some reason." Can anyone help me figure this out? I have such a hard time with regex and mod rewrite...

    Read the article

  • How to compile x64 asp.net website?

    - by Eran Betzalel
    I'm trying to compile (using Visual Studio) an ASP.Net website with the Chilkat library. The compilation fails due to this error: Could not load file or assembly 'ChilkatDotNet2, Version=9.0.8.0, Culture=neutral, PublicKeyToken=eb5fc1fc52ef09bd' or one of its dependencies. An attempt was made to load a program with an incorrect format. I've been told that this error occurs because of platform noncompliance. The weird thing is that although the compilation fails, the site works once accessed from a browser. My theory is that the IIS compilation uses csc.exe compiler from the Framework64 (64 bit) folder while the Visual Studio uses csc.exe compiler from the Framework (32 bit) folder. If this is acually it, how can I configure my Visual studio to run with the 64 bit compiler for ASP.Net sites? This is my current development configuration: Windows 7 (x64). Visual Studio 2008 Pro (x86 of course...). Chilkat library (x64) IIS/Asp.net (x64).

    Read the article

  • Replacing build.xml with Build.java - using Java and the Ant libraries as a build system

    - by Dean Schulze
    I've grown disillusioned with Groovy based alternatives to Ant. AntBuilder doesn't work from within Eclipse, the Groovy plugin for Eclipse is disappointing, and Gradle just isn't ready yet. The Ant documentation has a section titled "Using Ant Tasks Outside of Ant" which gives a teaser for how to use the Ant libraries from Java code. There's another example here: http://www.mail-archive.com/[email protected]/msg16310.html In theory it seems simple enough to replace build.xml with Build.java. The Ant documentation hints at some undocumented dependencies that I'll have to discover (undocumented from the point of view of using Ant from within Java). Given the level of disappointment with Ant scripting, I wonder why this hasn't been done before. Perhaps it has and isn't a good build system. Has anyone tried writing build files in Java using the Ant libraries?

    Read the article

< Previous Page | 35 36 37 38 39 40 41 42 43 44 45 46  | Next Page >