Munich Breaks Even By Migrating To GNU/Linux

Mayor Ude of Munich, Germany, has stated some facts about the effects of the migration from that other OS to GNU/Linux:

So much for the sycophants of M$ claiming costs had ballooned with GNU/Linux. Frankly, I am surprised they found so few problems with that other OS. Perhaps users “just rebooted” and made problems go away with that other OS. 46 per month with GNU/Linux is rather trivial for thousands of desktops. The help-desk people must nap a lot. I’ve had that many requests to reset passwords with ~100 PCs. I’ve had a few users who forgot passwords every weekend…

Good for the taxpayers and IT people of Munich. We know that cost was not the only consideration but it is a welcome side effect.

About Robert Pogson

I am a retired teacher in Canada. I taught in the subject areas where I have worked for almost forty years: maths, physics, chemistry and computers. I love hunting, fishing, picking berries and mushrooms, too.
This entry was posted in technology. Bookmark the permalink.

45 Responses to Munich Breaks Even By Migrating To GNU/Linux

  1. oldman says:

    “I am sorry to have to say this that you think that your environment is special then you are a idiot.”

    No my environment is not special, but neither are you a$$hole! You speak as if you have some absolute right to come into an environment and make it over in your image by playing the security trump card. It doesn’t work that way in reality, jack. And you have blown enough bushwah and wishful thinking as if it were gospel that I am becoming convinced that the other posters here who dismiss you as a fraud may be correct.

    Whatever. The bottom line is this sir. You can babble untill the cows come home about the evils/problems of closed vs. open source. The fact remains that they are not going away any time soon, and you ARE in the end making a living from them as well, despite all of your protestations.

  2. oiaohm says:

    oldman by the way over my history I have taken on lead developers over quality of work. You are presuming I am anti-microsoft you never bring me the questions about some of the secuirty pracs of particular distrobutions or other things. Yes when I review stuff I sort out where the vested interests are. You should see the things I have said about apparmor over the quality of secuirty that provides.

    Sorry to say oldman knowing who is working for you, who is working against you and who will change sides when they suite. TAM from Microsoft will change sides when it suites them. All is part of secuirty auditing and developing secure systems.

    So yes I have a vested interest in fault finding since that is my job. Does not matter if it FOSS or closed source. Yes performing risk assessments is also part of my job. From the way you talk its not yours oldman.

    Someone who job is to locate threats sees things far clearer than people who job is not.

    Heck I am currently battling with the posix standard body over particular things that have been causing on going problems in the hope of getting them cured once and for all. Oldman you are only seeing one side of me because you keep on setting yourself up to only see that side.

  3. oiaohm says:

    oldman that time it solved your problem. I have a vested interest from secuirty of knowing who is working for who. So its important to remember a TAM at Microsoft is not always working for you. Sometimes they are giving you stuff they want tested. So handle those exclusive hot-fixes careful. Because you are not in a position to see enough to assess what one is trust-able and what one is an experiment.

    “Get the best out of my staff? You havent a clue of my environment sir, nor do you ave a notion of the deliverables and the constraints. It is all fine and good to talk of theory, it is quite another to deal with reality.”
    This is where you are wrong. Deliverables and constraints I do have a very clear view of.

    I am sorry to have to say this that you think that your environment is special then you are a idiot. Reason same things are required so coders and software for the business developers effectively.

    What I am talking about is using what ever means is required to achieve the Deliverables and work around the constraints of the skill level of staff I have

    Expanding the skill and knowledge of staff is a critical process to future production being met and keeping up with competitors does not matter the field. There are some universal constants.

    Microsoft VAR registration is very handy when migrating. Basically you are that thick that you don’t see that “pure linux support services” includes migrations from Microsoft to Linux. Its not like a migration happens instantly.

    I also have items like registrations for solaris an aix as well.

  4. oldman says:

    “You raised a issue with a person who knows exactly how they operate.”

    To put it bluntly I dont believe you sir. You have a vested interested in finding fault, and as far as I can see finding that fault you will, especially if Microsoft is involved. The fact that I have reported remain. The hotfix solved our problem, the service was given and it was delivered in a manner that was timely, and that is all that counts to me.

    “This is what you have to get oldman I might be talking to what you think is idealistic. But what I am really talking about is what you require to get the best out of the staff you have in the most profitable way. ”

    Get the best out of my staff? You havent a clue of my environment sir, nor do you ave a notion of the deliverables and the constraints. It is all fine and good to talk of theory, it is quite another to deal with reality.

    The fact remains that while you talk a good game about the wonders of FOSS you remain a Microsoft VAR and I suspect will deliver services and support of microsoft products. If you are so sure of the superiority of FOSS, then why not abandon your Microsoft VAR status and work giving pure linux support services, eh Mr O?

  5. oiaohm says:

    The other advantage of having a trade system to rent your coders out to other companies is they get a broader ideas of how the others in the market place is doing things.

    Quality of coders work is linked to the coders experiences. Even if you are normally producing closed source code for your company being rented out at times good for the coders development so they don’t become stuck in a rut of the internal way of doing things.

    Google give there coders like 20 percent of there employed time to work on outside projects. Smaller companies cannot afford to give up wages like this.

    Yes oldman if your company is not doing something of the things I am talking about you end up with isolated coders lacking experience that do not produce as good as results as they could have with more broad experience. Getting the coders a broad experience can be a profitable operation.

    Isolated coders not up on current quality control methods are one of my big pay checks for being called in to audit to find where the network got breached to plug the leaks.

    This is what you have to get oldman I might be talking to what you think is idealistic. But what I am really talking about is what you require to get the best out of the staff you have in the most profitable way. Nothing is better than being paid to expand experience of your staff so they can produce better work in future.

  6. oiaohm says:

    oldman “And you know this how sir? have you ever been associated with any institution or company who paid for premier support. Have you ever worked with Microsoft TAM’s?”

    I use to be a guy you would employ to come in and audit of secuirty of networks and buildings. So yes still deal with about 15 companies paying Microsoft support. This is how come I am very aware how TAM operate because those business had security issues. These were raise with the TAM the patch was provided only made for that business by the documentation. Of course being secuirty I did check what the patches were doing. So I had 10-15 identical patches to the same problem that the TAMs was telling me was made for that business and that business only. So yes I know exactly how TAM’s operated because I was in the right place to prove their lies. Yes it also was good if I got like 15 complete different patches to the same secuirty issue you knew they were experimenting.

    Does this explain why I am such a technical ass hole. I don’t have your closed one company view oldman. I have more of a overview since of the work I have done. Since I have been contracted between many companies at the same time.

    Just because yours is not operating does not mean what I am talking about is not going on.

    If you wish to make progress and be as cost effective as possible. Using the best quality coders todo the job your company will be operating on a trade system. Now of course there are ones that don’t have a set up trade system. These have great deal of trouble getting patches up stream at times due to there coders not having the right skill set and having no way to trade for the right skill set.

    There are limited ways to work highly effective with FOSS. Different companies are at different levels of effectiveness.

    “it doesn’t even come close to the reality of FOSS collaboration”
    Question is oldman how effectively are you working to get what you need out of FOSS. I would guess not very well. Since I would bet you have a growing collection of private patches that become more and more costly to maintain.

    Yes I will give seeing all I described inside one company is rare but it does exist.

    The TAM idea I think you brought up because you suspected that was above me oldman sorry its not. You raised a issue with a person who knows exactly how they operate.

  7. oldman says:

    “Really a Microsoft TAM is really not working for you. They are working for Microsoft to improve Microsoft products to exploit you to test the alterations they make.”

    And you know this how sir? have you ever been associated with any institution or company who paid for premier support. Have you ever worked with Microsoft TAM’s?

    I am willing to bet that you haven’t.

    As far as all the rest of your wall of text “information” is concerned, it doesn’t even come close to the reality of FOSS collaboration That I have seen operating in that side of our shop that does deal extensively with FOSS. In fac tit sounds suspiciously like more what you hope will be true than what is true.

    spare me the baloney sir.

  8. oiaohm says:

    aardvark lot of SME pay for redhat contracts. Fail to read the fine print so fail to use the number of coding hours they get in the contract. Yes these do add up year to year. One company I found had over 200+ coder hours at redhat to allocate to a problem and they were wanting to go out and employ someone random for 4 weeks. Simple call to redhat got job done perfectly.

    So yes some cases they have paid all the money they need to get the issue annoying them fixed. Just they have not sent a message to redhat of please use my coder hours to fix X or make my Y. Yes this was a complete custom website made by redhat not for release open source. Yes those coder hours can be used how ever you want just like if they were your staff.

    Scale defines method. Coders time is for sale in the FOSS world. Its just knowing who to buy it from and if you can sell as well to reduce the cost.

  9. oiaohm says:

    aardvark Really you get it wrong. You are forgetting a good coders time is a sellable asset.

    “I have to determine what skill-set is required to fix it. I have to contact an employment agency. I have to draw up a laundry list of requirements to match the skill-set. I have to rely on the employment agency to filter candidates down to a sensible three or four. I have to interview those three or four. I have to choose one, who might not be the one I want, which means I then have to rinse, lather and repeat. I have to provide office space and associated assets for the Chosen One.”

    Linux Foundation does not charge a fee for staff placement. Neither does any of the FOSS staff sourcing. They are paid by big enterprise to be looking out for skilled FOSS personal and get them placed somewhere so there resources will be around to assist. Small enterprise get good FOSS staff with no fees.

    Normally you don’t go to a normal employment agency with FOSS. Instead you contact a specialist like the Linux Foundation. Tell them what you need. The look at people working on those project the quality of there prior work and how much you are willing to pay. So find you a pre-tested coder who can work with normally 12 months pre-tested work. Yes there is more to the open source revision system to pull out what coders have done then just bug fixing.

    The summer of code each year is about finding new coders with skill.

    FOSS using companies normally get wise. They will never have all the developers in the areas they need. So you sell and buy coding time. Ie my coder will fix X bug that means nothing to us if you will fix Y bug that is important to us.

    If you don’t have a coder to trade you pay cash to likes of Redhat and IBM to use their coders.

    This is the big thing there is a commercial model behind FOSS. aardvark. The trade item is Programmer time with Programmer skill set. This is either traded barter or for cash. This trade is why FOSS is not free. You need resources.

    When I say employ a coder to fix it. I one of the following. Employ them directly, Barter with their boss to let one of my coders do something for them in exchange for them fixing something for me Or paying that coders boss to have them do it for me.

    Contractor for four weeks in FOSS world you would ring the likes IBM, Redhat, codeweavers and Orcale and commission the job to them. They will give you direct contact to one of there coders for the right money of course and these are tested quality coders. You would not even bother going through interview process. Job less than 6 months not worth the effort.

    Interview process is only if you are after a coder to be full-time staff for at least 6 months to provide you with time to sell or barter as well as do the work you require. Yes it is possible to come out with a profit from your on staff coders. Yes coders are a possible income stream.

    This is FOSS. When it don’t work you have the coder who got it wrong ass to kick.

    Yes redhat does not make 1 billion dollars of income a year for no reason. They have a good set of coders who time they can sell. FOSS does not sell the program we sell the developers time.

    Yes it is also possible to work as a collective to employ coders in the FOSS world.

  10. kozmcrae says:

    aardvark said:

    “Look, I don’t doubt that Seaboard Corporation can afford any number of useless dingbats on the company payroll.”

    Wow aardvark, you really hold yourself in high esteem. Your opinion really matters, to you. In fact, your opinion matters so much to you, it leaves no room for it to matter to anyone else.

    So even though you can type 120 wpm, you can leave out your opinions because no one else will bother to read them.

  11. aardvark says:

    So, FLOSS is aimed fair and square at companies with 10K employees now, is it? It’s such a shame when idealism shrivels and dies.

    Look, I don’t doubt that Seaboard Corporation can afford any number of useless dingbats on the company payroll. Some of those useless dingbats (probably more, in fact) will be employed to tidy up the mess that M$ leaves behind.

    But that doesn’t really change the obvious fact that an SME (conventionally 50 to 500 or so employees) that is not directly involved in the software business is not going to be impressed by Mr Oiaohm’s bizarre cost-benefit analysis.

    Oh, and I used the example of hiring a contractor for four weeks purely as a “best case” scenario. Hiring somebody on a permanent basis for something that can be fixed via an annual contract at $5000 or thereabouts is so utterly silly as to be equivalent to walking into the boss’s office and asking to be fired on the spot.

  12. aardvark wrote, “I have to analyse my problem (luckily, I can see all the thousands of lines of badly-written code). I have to determine what skill-set is required to fix it. I have to contact an employment agency.”

    That’s reactive thinking. FLOSS is all about proactive thinking. Having someone on staff in charge of local implementation of a FLOSS project and contributing bug-reports and code-changes upstream is affordable, efficient and proactive, not reactive. You want to be in a position to install features and fixes so that if a problem arises there is someone up to speed right away. Ideally, this person would be the Maytag Repairman of the company, but it’s still cheaper than buying licences if you are a large organization. By contributing stuff upstream, you get easier migration to later releases because the fixes you need are in the package-managed releases from upstream and the world is a better place. It’s a matter of scale. If you have 1000 employees and 1 of them is a programmer minding LibreOffice, it may not be efficient to pay for software that way, but if you have 10K employees, it’s a trivial decision to make.

    Seaboard Corporation:“Seaboard Corporation directly or indirectly employs more than 14,000 people worldwide. We are #500 on the 2011 Fortune 1000 list with net sales of approximately $4.4 billion annually. Seaboard is traded on the NYSE AMEX Equities under the symbol SEB.”

    They can easily afford one or more FLOSS programmers in lieu of paying for licences.

    Also, “According to analysts, we have just passed the open source code’s first wave, mainly “early adapters”, the pioneers, it happened on the train. We are now entering the second wave, when the large mainstream companies driving this trend.

    – We advise CIOs to today to abandon the image of open source as “anti-commercial” and attt begin to sketch out a corporate policy for open source, says Nikos Drakos.”
    see Open source is gaining ground among enterprises

  13. aardvark says:

    Oh, so that’s what a TAM is. I’ve never met one.

    “FOSS you want something fixed you employ coder to fix it then they are really your staff and you really do have a say that they do fix what you want.”

    What a wise way to go about things. Let’s see. I have to analyse my problem (luckily, I can see all the thousands of lines of badly-written code). I have to determine what skill-set is required to fix it. I have to contact an employment agency. I have to draw up a laundry list of requirements to match the skill-set. I have to rely on the employment agency to filter candidates down to a sensible three or four. I have to interview those three or four. I have to choose one, who might not be the one I want, which means I then have to rinse, lather and repeat. I have to provide office space and associated assets for the Chosen One. I have to pay him or her $4000 (assuming a four week contract at a derisory $200 a day), and I have to pay the employment agency $1000 on top of that.

    After all that, I get the same junky software with a badly-tested kloodge hanging off the side, attached by duct-tape and a set of PHP and bash scripts.

    Yes, that would be the way to go. Hang the efficiency of a centrally-managed support system that solves each problem for all affected customers, priced on an up-front contractual basis with no hidden extras. That would just be silly.

    FOSS wins again!

  14. oiaohm says:

    aardvark
    “What’s a Microsoft TAM?”

    Microsoft Technical Account Manager. This is the microsoft title. Open Source world these would be a Bugzilla Triage personal or what Redhat call BugZappers.

    There job in reality is part evil. They talk to developers and get patches for customers in the Microsoft sense since you have to pay to have access to a TAM.

    Microsoft TAM will tell you that you have been given an exclusive patch to a problem that is a flat face lie. I know this because I have worked as consultant between companies so I have seen the MS TAM do this. Two companies both got what I called unstable hotfixs yet told they were the only one to get it there were binary the same file. Works to keep the hotfix secret as well.

    Other than lieing that Bugzilla Triage people don’t do. There main objectives are:
    1) To locate duplicate reports and manage them effectively.
    2) report and attempt to get fixes to bugs that are not duplicate.
    3) get test results back from users that the patch works so you don’t need testers.

    Microsoft TAM also do getting documentation as well. Area that FOSS Triage people could improve on.

    Really a Microsoft TAM is really not working for you. They are working for Microsoft to improve Microsoft products to exploit you to test the alterations they make.

    And just like a bugzilla Triage people Microsoft TAM personal have no requirement to fix your problem. They are not your staff.

    This is why I cannot understand why companies want to pay for this in lots of cases.

    FOSS you want something fixed you employ coder to fix it then they are really your staff and you really do have a say that they do fix what you want.

    Yes you find a lot of open source project bugzilla triage people turn out to be developers employed by other companies who want the same bug as you fixed and are looking at you as a person to be the guniea pig for the fix. But at least you have not paid to be abused in FOSS way. Being a guniea pig the FOSS way is a very cheap way to pay for work to get done.

    MS you are paying Premier support to be treated like a guniea pig.

    Redhat in your support contract is so many hours of real developer time you can allocate to any problem you like. So again you are in control not a guniea pig lining up to be abused.

  15. oiaohm says:

    aardvark 10 year migration not a day of downtime to required operations happened in that complete time.

    Yes its possible todo conversions faster but you do risk downtime.

    aardvark really don’t try to create words. Data entry operators I have more respect for than testers. At least they are feeding in real world data that normally cannot be replicated by a computer program.

    Testing require repeatability. Testing is a job for machines so it can be performed regular enough.

    Users reporting bugs is far more useful than human drone testers.

    “And I assume they have been paying for Windows licenses” purchases of them have been included in the cost reduction maths.

    Really the Munich project breaks in two. 5 years sorting up mess of office processing systems on Windows. The start of the Linux full roll out was delayed until 2009 due to how many cases of departments using duplicate templates and documents without need and macros and other lets clean the mess out our offices.

    900 macros become 100 macros todo all of the same jobs. This tells you something reinventing the wheel was going on a lot internally at Munich. More you have reduces means to audit if the macros are safe. Munich problem was a secuirty issue that required fixing. Cost time.

    The migration to Linux has been quite fast really. Migration to Openoffice on Windows was also quite fast at only 5 years.

    Remember it has not blown it budget it come out underbudget and overtime. Now this is a oddity. Over time normally equals over budget as well.

    Remember when Munich started the forecast was they would not make it at all.

    Sorry I have done Microsoft migrations 3-5 years is not on common. That the OS change and the Office suite change was done in Munich as two independent stages makes the 10 years acceptable.

    How you get to 3-5 with the Microsoft conversion is OS and Office suite change at the same time.

    Really did you not read the Munich roadmap and go hmm if I did this with a MS change over how much time would it take. The answer is 10 years+. So your failure claim is bogus.

    French police did a OS and Office suite change at the same time and pulled it off for there complete area from Microsoft windows and office to Linux and OpenOffice in 2 years. Different roadmap different speed.

    Both at same time 3-5 years normal. One at a time 6 to 10 years is normal.

  16. Kozmcrae says:

    aardvark says nothing: Ends with saying:

    “Do you want a diseased, ten-year-old, failed project where everybody from the politicians down are making excuses?”

    That’s just a bunch of whistling in the dark aardvark. You are having trouble accepting Munich’s success so you spew a bunch of crap that doesn’t say anything, isn’t relevant to anything and has no bearing on the outcome.

    You have no idea how foolish you look right now. Good thing you are clueless.

  17. aardvark says:

    Changing 900 of anything does not take ten years, Mr Pogson. Otherwise we would all be stuck in the Stone Age.

    It doesn’t matter what you or I think of the Munich experiment, and it frankly doesn’t matter which OS they chose to piddle around on.

    The fact is that ten years for a user base of 15,000 is an abject failure.

    Celebrate it if you want.

  18. Changing 900 of anything is not trivial. Munich’s highest priority was not to break anything. That requires a lot of study before a change can be made. To do it system-wide with people in diverse offices all over a city is a further complication. It would take way more than two people because it was not just a job for technologists but users and managers from separate groups. e.g. suppose there was an “addition macro”, something about making an entry into the spreadsheet or database. The technologist may well see that as something that must be standardized and he would scour the lists looking for macros that “added” and produce one that did the same thing all over the city. That would be rational but frightening to all the stake-holders who have to check the type of source-data and the type of target-data and test the operation on their data to make sure stuff is not messed up. If that’s not enough, they might have to work around M$’s weird ideas about dates or other values. I can see a dozen people fussing over that one. Remember there are 11 units to the organization and they may well have quite different data to work on. Presumably they have to merge databases, too, not just the macros. Then, some macros have to be eliminated. That can be tough, especially if A likes the thing and B does not. There’s a lot of give and take, not just logic and arithmetic. Further, all dependencies have to be found for all the macros, a mechanism that was not likely in place when the migration started. So, there are several layers to the problem of rationalizing the macros, fraught with difficulty like any merger. Mergers of large organizations often take years to complete. It’s a one-time pain for long-term gain.

    Even if they had kept the M$ environment, they may well have had similar work to do moving from NT4 to XP and then to “7”. How many times did M$ change file-formats, user-interfaces and APIs in the process? How many versions of Excel did they have in the process which Munich would have to test each time? A lot of the rationalizing was something they needed to do in any event.

  19. aardvark says:

    Mr Oiaohm:

    Thrilling, Munich, isn’t it? Let me quote:

    “Originally there were approximately 900 uncontrolled and undocumented macros. We have replaced these by 100 macro applications in our orderly, documented and quality-controlled “macro repository,” and by 38 centrally maintained Web applications.
    We achieved a total consolidation rate In the conversion of Office templates and forms of approx. 40%.”

    900 uncontrolled macros, and ten years. Let me just take a punt at how long it would take to do that in, oh, I dunno, a Microsoft environment?

    Three people, two years, max.

    Y’see, this isn’t about magic pixie dust. It’s about requirements.

    And they’re still stuck on OOo.

    And they’ve got problems hiring drones (I know how much Oiaohm loves drones: it’s the way he thinks of testers. Imagine how much loathing he must feel for mere data entry clerks).

    And I assume they have been paying for Windows licenses (and Office licenses) for the past ten years for the non-converted computers — there were rather a lot of them, for ten years — and I assume they are still paying for the bosses’ machines, which no doubt will not rely on broken technology. (“Well, I would have, but you know how it is. Gotta communicate with the guys in Berlin, y’know?)

    Just as an interesting side-note: you do realise that 20% of the so-called “consolidation” was simply to reduce the number of me-signy things, don’t you?

    It’s beyond pitiful. It’s farcical.

    And perhaps more important, it’s a very good lesson to any other provincial government. Do you want a diseased, ten-year-old, failed project where everybody from the politicians down are making excuses?

    Yup. Good marketing slogan there.

  20. aardvark says:

    Mr Oiaohm:

    What’s a Microsoft TAM?

  21. aardvark says:

    Well, at least I’m the family troll.

    Don’t drive over any nearby bridges over the Easter week-end, y’hear?

    I’m privy to secret Troll information.

  22. aardvark, the troll, reviled LibreOffice’s work with comments like, ” Libre publishes no such metrics. Since there are no metrics, there is no poor quality work. “

    Clearly aardvark either has no concept of how FLOSS works or how code is developed. One looks at individual bugs and fixes them on the one hand and counting bugs is quite another matter. Debian GNU/Linux, for example, does count bugs and out of 30K+ packages is working on a list of a few hundred bugs, LibreOffice’s included. That is great performance, bug-wise.

    Of course, LibreOffice counts bugs. Searching for “open” bugs finds 5K of them here. Restricting the search to “verified” bugs finds 27 actual bugs. That is wonderful code quality. The product clearly works for me. The few bugs I have found are mainly wishlist items not worth reporting, such as default font sizes for my poor old eyes…

  23. oiaohm says:

    oldman
    “I really have to tell that one to our microsoft TAM – he should get a good laugh! Premier support has worked with us on several occasions to troubleshoot several arcane bugs.”
    Go ahead. Learn that you are a twit as he laughs at you for thinking that you have any form of support. You are free testing staff for issues in there code base who pay for the right. They have no requirement in fact to assist you at all if they decide it too hard to fix. Having Premier support means you will get so much but its not insurance that they will fix it. This is reality and you wonder why I call you a idiot.

    You have faith in something because it has provided you with results when it suits Microsoft. Not based if it suits you.

    “even got us our own hotfix AND they stayed with us until we verified it fixed the problem.”

    Interesting right. They did not use there own testers since paying testers costs money they used you Oldman. You were the guinea pig. I would prefer not at any times to have my system as the guinea pig for some of the experimental code.

    FOSS projects you get to avoid using commercial items to quality test. You get to choose people who are willing guinea pigs to test code.

    “The problem for you is that all of our major ERP systems are licensed under source code available licenses.”
    See you have to use FOSS. Do you have any upstream control so that repaired bugs don’t come back. I guess you don’t submit upstream properly.

    “Of course we pay for that support, as we do for our RedHat support. We have a throat to choke, unlike you.”
    Having a throat to choke does not bring back someone life. Yes where I am does pay for support were suitable. Reality is reality. You are avoiding having to set up proper quality metrics since if something goes wrong we will blame Redhat or Microsoft basically anyone bar one selves for not having good enough quality control processes.

    There are metrics on Libreoffice.

    http://wiki.documentfoundation.org/QA/Testing/Automated_Tests there are Automated and Manual testing there is even a buildbot. So you don’t have to trust the metrics they provide. Instead you can run the stuff and generate your own metrics even add you own to the tests. So that a regression that brings back a defect does not cause your company trouble.

    Its grammar checker not spell checker yes I know the one you are referring to lightproof and its not just Hungarian.

    If you are referencing to Hunspell that predate Libreoffice existence and was done to match sun micros will for closed source package.

    Libreoffice is far more on the path to becoming correct than OpenOffice ever was.

    If you are depending on something you want automated testing. Since this make sure what you think is tested is.

    Basically aardvark you did not check out Libreoffice what they are doing so are a twit.

  24. aardvark says:

    Mr Oiaohm:

    Apologies for not applying myself to your latest absurdist effort; it’s a little hard to keep up. I need to rebuke Mr Oldman first:

    Mr Oioaohm: “If your coders are working on lets say libreoffice with a requirement to submit all alterations upstream. Libreoffice core provides the project management of the developers and testing. Poor quality work does not make it. So your requirement is that X task gets done to project acceptable quality. Boy that kills of having manager do much other bug the developer is it there yet.”

    Mr @ldman: “Working on a desktop app? surely you jest. Our our development teams are working on tailoring commercial ERP systems in the Peoplesoft class to meet business needs.”

    Like many commenters on this blog, I am desperate to apply recognised standards: you will note that I have accepted the six-year-old retarded insults of Mr McCrae as a standard moniker for Mr @ldman. In fact, it looks rather distinguished. But maybe that’s just me.

    Anyway, to the rebuke: Atman, you old fool, you’re missing the Ocker’s point. Easy to do, I admit. Concentrate while you’re slumping back there in your bath chair, will you?

    Mr Oiaohm is specifically referring to a single project. Not your project: Libre @ffice. (Sorry, my keyboard got stuck.) He mentions:

    “… [the] requirement to submit all alterations upstream.” A beautiful dream, isn’t it? Unfortunately, as with dreams, you wake up and the so-called requirement never actually happened in the first place.

    “Libreoffice core provides the project management of the developers and testing.” Another beautiful thought. Is it documented anywhere? I have this glorious vision of teenagers trying to work out what a spline curve is, whilst creating a totally new spell-check regex system in Hungarian. (I’m not joking: that’s what Libre did.) And of course a bunch of pigeons with PhDs who will peck at buttons in the Oiaohm-approved manner of testers.

    “Poor quality work does not make it.” Because, you know, you need actual metrics to define what is “poor quality.” Libre publishes no such metrics. Since there are no metrics, there is no poor quality work. Simples!

    “So your requirement is that X task gets done to project acceptable quality.” Bad, Bad Mr @ldman! What an absurd requirement!

    It’s Z task, not X task, you demented half-w00t!

    “Boy that kills of having manager do much other bug the developer is it there yet.”

    OK, I can’t make head n@r tail of that. I’m sure it’s slightly profound, though.

    Mr Oiaohm: Apologies once again for not dealing with your latest Roll of Wisdom. I’m feeling slightly constipated after all that effort. I may use it to wipe my backside with tomorrow.

  25. oldman says:

    “Read Microsoft license you have no requirement that they fix anything other than good will either.”

    I really have to tell that one to our microsoft TAM – he should get a good laugh! Premier support has worked with us on several occasions to troubleshoot several arcane bugs. One of them even got us our own hotfix AND they stayed with us until we verified it fixed the problem.
    Of course we pay for that support, as we do for our RedHat support. We have a throat to choke, unlike you.

    “Only way to make fully audited and quality code is to have the full source code. If you don’t have the full source code and you are wanting quality you are dreaming.”

    The problem for you is that all of our major ERP systems are licensed under source code available licenses. It turns out that the levels of tailoring and customization that we do require having source code.

    all the rest of your wall of text is as far as I am concerned bushwah. I suspect that I could drive a truck through what you don’t know sir.

  26. oiaohm says:

    oldman
    “The problem is those other parties that you would have us rely on have a) no credentials that are verifiable or b) no responsibility beyond good will to provide timely TESTED fixes to any of their code. furthermore they are not even responsible for their code!”
    Read Microsoft license you have no requirement that they fix anything other than good will either.

    oldman
    “We are not building some piddly crap add on to some piece of badly designed FOSS crap code that someone found on the internet.”

    This is exactly the answer you find from a closed source idiot. This means you are not working with the upstream. Random-ally found from the internet is not the case.

    To get good code out of FOSS does require working with the projects a little. Like assisting them to run build bots and testsuites(this is a key bit). Reason this way when you do get a fix done up stream it stays fixed. So the quality of the code base you are depending on then just starts increasing and increasing less work. This is why internal patching is bad. Like with android the most defective code in android was the stuff google developed internally without peer review. Higher defect rate than the rest of the Linux Kernel. Secuirty grounds it has to merge back in.

    Simple point here oldman you don’t work in a productive relationship with FOSS. Instead you just take from it. Just taking equals the application kicking you in your teeth so you start to believe it bad code. Not that the fact the FOSS code is kicking you in the teeth over and over again is part your fault not working with the upstream to put the systems in place so bugs once killed never can come back.

    oldman
    “Our our development teams are working on tailoring commercial ERP systems in the Peoplesoft class to meet business needs.”
    So you are going to be under reviewed.

    aardvark
    “Testers, outside whatever goes on in your mind, are emphatically not second-class citizens and button-pressing monkeys.”
    In fact most testers are in fact in the class of button-pressing monkeys as time gone by. Your own internal testers are worthless in most cases. Why because they become button-pressing monkeys who do the tests exactly the same way. So missing many bugs. You want random source of guinee pigs who will try the application to see what features are coming and understand that it could fail. This is where FOSS is great.

    The button-pressing monkey stuff should be in test-suites automated and added to build bots so every patch alteration gets checked by test-suites. This will be helped by klee.llvm.org that tells you if you testsuite covers every single code path.

    Wait oldman you don’t have the source code to the ERP you are working on. So stiff you can never be 100 percent sure that you testsuite has 100 percent coverage or that you coders are testing everything so you code is crap. That is the problem people like oldman try to make out that crap code is good quality when they really don’t have the documents to prove it.

    Yes the argument that closed is better than FOSS is a lie because closed developers normally never can generate the complete top to bottom audit report.

    Labour wise running testsuites cost bugger all even power wise cost bugger all. Running items like klee costs days.

    Secure development testers what are they. They don’t exist. Testsuite checking conformance replaced them years ago.

    Only way to make fully audited and quality code is to have the full source code. If you don’t have the full source code and you are wanting quality you are dreaming.

    Oldman I from time to time do systems that if they malfunction will kill someone not just make someone broke like the ones you are doing. So I serous-ally know what quality code is and what is required to make it. If someone said to me someone’s life depends on this closed source bit of code I cannot audit I would tell them that its not going in unless a NDA can be got giving us access to the source code.

  27. oldman says:

    “@ldman yip yap yap yap …… to.”

  28. Kozmcrae says:

    “Thank you Mr. A. The truth is that I have done just enough software development in my day and dealt with FOSS and its predecessors long enough to know where its bodies are buried, so to speak.”

    @ldman finds a replacement for his well worn Love Glove. I’m sure you two will spend many hours together stroking each other’s egos. Someone has to.

  29. oldman says:

    “Mr Oldman quite obviously knows more about actual software development than you do. ”

    Thank you Mr. A. The truth is that I have done just enough software development in my day and dealt with FOSS and its predecessors long enough to know where its bodies are buried, so to speak.

    As you may probably know if you have been reading this blog Mr. Oiaohm and I have crossed verbal swords quite a bit in the past. I have become immune to his gratuitous insults and bullying posts and have no problem telling him where to stick them.

    He does however if only by accident bring up points worth commenting on, if only to counter them.

  30. oldman says:

    “This is why you want code shared with other parties so it can be reviewed and poor quality solutions get pulled up.”

    The problem is those other parties that you would have us rely on have a) no credentials that are verifiable or b) no responsibility beyond good will to provide timely TESTED fixes to any of their code. furthermore they are not even responsible for their code!

    “That is because you are a idiot only a idiot would suggest the idea that having developers meant they have to be managed by you fully this is the closed source way when using FOSS. Closed source way is insanely expensive because you don’t exploit the free resources out there.”

    And you are qualified how to say this sir? As far as I can determine from your words in previous post you are nothing more than an itinerant PC tech providing ad hoc IT service to remote communities. That is all I know about you sir. For all I know you are just blowing it out your backside about the rest.

    “If your coders are working on lets say libreoffice with a requirement to submit all alterations upstream. Libreoffice core provides the project management of the developers and testing. Poor quality work does not make it. So your requirement is that X task gets done to project acceptable quality. Boy that kills of having manager do much other bug the developer is it there yet.”

    Working on a desktop app? surely you jest. Our our development teams are working on tailoring commercial ERP systems in the Peoplesoft class to meet business needs.

    ” Cost of testers is a zero factor. There are tones of people who are not your company who will test your software for you because they are feature greedy and want to try the latest. ”

    ROFLMAO!!!! Your really dont have a clue about how true IT works do you Mr. Microsoft VAR! NOBODY is going to test our custom code for us but us. We are not building some piddly crap add on to some piece of badly designed FOSS crap code that someone found on the internet. Deliverables have to be met, and you dont do it in the “no throat to choke” world of the Million man random bug generator that is FOSS.
    “Oldman how much internal crap code do you have. If its not good enough to sell or release FOSS you should really start thinking should we even have this program.”

    So now you are a software development expert eh, Mr. Microsoft VAR. You life must be very complicated having to manage all those projects while delivering Itinerant IT in the Australian OutBack!
    I think I can safely assume from your comments that you don’t even have a clue, Mr. Microsoft VAR.

    Spare me the wall of bushwah, and your insults.

  31. aardvark says:

    Mr Oiaohm:

    I’m trying to be as polite as possible here, so forgive me if this sounds abrupt, but it has to be said:

    Please leave the business of IT management to people who know what they are doing. You’re most welcome to contribute (FOSS, proprietary, or otherwise) as an individual developer, but my goodness what a thorough-going shambling disaster you would be in any sort of managerial role, one where you actually get to make the decisions.

    Just to take one particularly unfortunate thought:

    “Having testers for code development your a foss using company are you nuts?? Testers are only for testing deployment images as per closed source usage.”

    This would be quite bad enough without your SOP throwaway insult. I can think of a dozen different areas where testers are invaluable, quite apart from pointing the latest distro at a CD burner. Requirements specification; customer liaison; documentation; process quality assurance (useful even for minimalist agile development), many more.

    Testers, outside whatever goes on in your mind, are emphatically not second-class citizens and button-pressing monkeys.

    Oh, and stop suggesting that your betters are “nuts” or “morons.” Mr Oldman quite obviously knows more about actual software development than you do. For that matter, so does everybody on this site … and I mean everybody, with the probable but not guaranteed exception of Mr McCrae.

  32. oiaohm says:

    oldman Little truth of the matter. Dirty internal only hacks done with closed or open source software will cost you money. Munich messy template and macro system was costing them a fortune.

    This is why you want code shared with other parties so it can be reviewed and poor quality solutions get pulled up.

    “that any savings are dwarfed by the increased costs of have to build and manage a team of developers Project managers and testers.”

    That is because you are a idiot only a idiot would suggest the idea that having developers meant they have to be managed by you fully this is the closed source way when using FOSS. Closed source way is insanely expensive because you don’t exploit the free resources out there.

    If your coders are working on lets say libreoffice with a requirement to submit all alterations upstream. Libreoffice core provides the project management of the developers and testing. Poor quality work does not make it. So your requirement is that X task gets done to project acceptable quality. Boy that kills of having manager do much other bug the developer is it there yet.

    In fact when you don’t have stuff for your coders todo you can rent them out of course with limitations on what license code they can release as so your company can gain from others.

    When you deploy solutions you have to go through quality assurance. This does not matter if it closed or open. Cost of testers is a zero factor. There are tones of people who are not your company who will test your software for you because they are feature greedy and want to try the latest. Paying your developers to work on automated test-suites for critical parts is worth while investment.

    Having testers for code development your a foss using company are you nuts?? Testers are only for testing deployment images as per closed source usage.

    Yes as a commercial you view some of the FOSS world as guinea pigs to test out new features before using it in your production environments.

    Lot of companies develop some in house only applications. These are normally bug ridden. Yes normally if its not good enough to release that another company someone else on this earth cannot take advantage of it your have done something seriously wrong with your in house code base.

    Oldman how much internal crap code do you have. If its not good enough to sell or release FOSS you should really start thinking should we even have this program.

  33. oldman wrote, “Now multiply that effort by multiple applications and you get the dirty little secret of FOSS – that any savings are dwarfed by the increased costs of have to build and manage a team of developers Project managers and testers”.

    Not bloody likely. e.g. LibreOffice has hundreds of developers working for tens of millions of installations. Supporting those developers is much cheaper than paying M$’s licensing fees. Smaller projects need only a few developers and only a few businesses contributing. The world can and does make its own software. The world does not need M$.

    Consider the world’s PCs. There are about 1500 million PCs. If each of them contributed $50 towards M$ that’s $75billion. GNU/Linux costs much less to produce, distribute and support. Instead of griping that GNU/Linux is not $0, oldman should notice that M$ and partners are way over-priced.

  34. oldman says:

    “Those developers in house do not normally cost more if you allow them to share code with other companies as well. So sharing the work load. Yes if you have 7 developers in house and don’t allow them to share code with others and be reviewed by others they can be expensive.”

    What are they going to share? Local custom hacks? and what are they going to get from the outside? if they are lucky more base code that MAY be useful as is, even assuming that it is worth using. Now multiply that effort by multiple applications and you get the dirty little secret of FOSS – that any savings are dwarfed by the increased costs of have to build and manage a team of developers Project managers and testers, all where no such investment was needed.

  35. oiaohm says:

    Munich in the past 8 years with LiMux. Has fully changed distribution twice. Over all done 5 full distribution upgrades.

    aardvark also LiMux is not a homebrew project. Its a custom install disc based off of either Debian or Ubuntu. They don’t maintain there own independent distribution other than linux equal to a windows slipstreamed install disk. The projects that failed to migrate have been homebrew attempts. Project that pass are custom installers based off some other distribution.

    Pays todo you homework before making stupid comments aardvark you statement is false.

    http://linuxcoe.sourceforge.net/ makes the process fairly much like stamping a template on new distribution for a new release or distribution change. So very labour low.

    Custom installer disk pays for it development time since its development time is mostly bugger all.

    oldman long run no MS will cost you more if you mandate staying current. Munich has pulled of in place upgrade/replacement every 2 years.

    Those developers in house do not normally cost more if you allow them to share code with other companies as well. So sharing the work load. Yes if you have 7 developers in house and don’t allow them to share code with others and be reviewed by others they can be expensive.

    From a secuirty point of view Munich is better with FLOSS as well since they are upgrading way more often.

  36. Nope. “apt-get update;apt-get upgrade” will get them well on the way to their next roll-out. For a school with 100 seats that takes just a few minutes. Even for Munich, they can likely do something in a similar length of time, rolling out new images or installing new packages. It all works. This is one of the huge strengths of the GNU/Linux system with a good package manager such as Debian uses.

  37. aardvark says:

    Mr Pogson:

    “Frankly, I am surprised they found so few problems with that other OS [elided]… 46 per month with GNU/Linux is rather trivial for thousands of desktops.”

    Numbers are good. Numbers are essential. I’m glad to find out that there is a range, in this case, of customer tickets between that which is “typical” and that which is “unacceptable.”

    Apparently, both numbers are surprising, but I’m going to go with your methodology (as expressed a couple of posts ago) and admit that one or two wonky data points do not invalidate the data set as a whole. Not even if those numbers are quoted off an autocue by a regional politician with a lot to lose, rather than by an external auditor of some kind.

    The numbers here are therefore: 46 Good, 70 Bad.

    I’m not sure where the “tipping point” is, in your mind. I’m guessing it’s nearer to 46 than it is to 70.

    Either way, and as you are honest enough to point out, this workload would hardly tax a single support employee.

    Whereas ten years of pointless thrashing around with a mixed environment of heterogeneous machines, one set of which is built off a home-brew project for no clear reason (isn’t Red Hat or even Debian or Die Deutsche SuSE good enough for these lunatics?) … well, that must have cost a whole heck of a lot more than a single member of staff in the support team. At an educated guess, it would have cost roughly twice as much, for ten years, as it would have done to support one or the other systems.

    I love this “rolling savings” argument, btw. You don’t even have to look at it very closely to recognise that it is utter bollocks. Ten years ago, “small smart thingies” were not even on the horizon — so Munich has spent ten whole years (in perspective, 15% of the entire existence of computers) mutating from pre-XP NT to an equally antediluvian and sui generis fork of Ubuntu. And now they’re presumably going to spend another ten years catching up with 2011.

    It’s all good.

  38. oldman, referring to manpower to run GNU/Linux, wrote, “Who will cost far far more than the microsoft licenses sir in the long run.”

    Not at all. Gartner reported people are paying $2K per annum per PC to keep that other OS running. I can run a PC for a lifetime for less than that using GNU/Linux.

    1996“An oft-cited Gartner Group report estimates that companies must spend more than $6,000 a year on hardware and software upkeep for a $2.000 PC that runs Windows 95.”

    By being smarter, Gartner stated costs had fallen to $800 by 2008.

    Yet, one can have a thin client run fanlessly for a decade and get RedHat to manage the software on the server for peanuts ($800 per year for a server that can run hundreds of client PCs or $49 to $300 per client for various support options). Many organizations using GNU/Linux spend less than $100 per annum per PC.

  39. The savings are much more than licensing costs. With that other OS, Munich would likely have replaced OS and PCs twice over by now. It has been 8 years since this migration began. The fact is Wintel depends on the treadmill running forever, so the cost of sticking with Wintel is infinitely higher than the cost of sticking with GNU/Linux if one only considers licensing. Wintel is higher by a factor of two or three times if licensing and hardware are considered. Wintel is higher by an even larger factor if malware, re-re-reboots and manpower are considered. With the centralized scheme Munich has developed they should be able to dramatically reduce manpower but they have not as they used manpower to improve the whole system not just to migrate to GNU/Linux or to hold the hands of that frail OS from M$.

  40. iLia says:

    And it seams to me that the cost of licenses for MS Office is included:

    In calculating the costs saved from Munich’s mayor Christian Ude of 15,000 Microsoft Office licenses and 7,500 Microsoft Windows licenses that need to be partly purchased.

    this is a citation from the google translation of the original article so the cots of MS Office is included in the savings!

  41. iLia says:

    Upps, I wanted to write that the salary of a German clerk is $ 2,351 and savings is only $ 23 per one computer it is almost nothing.

    And now I am going to say something very politically INcorrect. If the government of Germany wants to save a lot of money they can simply stop paying the huge amount of money to this people who came from foreigns parts to Germany, do not work, and abuse the local people in many ways. And this will save much more money and lives as well, simply because many potential criminals will have to go home.

  42. oldman says:

    “Once you add in the savings of OpenOffice they can employ quite a large team forever more due to no longer paying Microsoft.”

    Who will cost far far more than the microsoft licenses sir in the long run.

  43. oiaohm says:

    iLia key word is 4 million so far. So the saving don’t stop at 4 million.

    15.52 million is the cost to return to Windows just for 10,000 machines. So I would say returning to windows cannot be considered.

    Every 3 to 4 years there is another 2.8 million in licensing costs without upgrading hardware staying with windows on 10,000 machines.

    2.8/4 is quite a decent figure. 700,000 a year. That is a few extra staff a year. 7 developers at a 100 thousand per year to develop there own software does not sound bad.

    Migration saving cost is not much. The ongoing is where the cream is.

    Remember this saving cost is Linux vs Windows.

    The cost saving of migration to OpenOffice vs MS Office is not in these figures. Yes this is more money.

    Once you add in the savings of OpenOffice they can employ quite a large team forever more due to no longer paying Microsoft.

  44. No government should ever neglect a million of anything.

    The prime motivator for the migration was not cost. That is just a fringe-benefit. The motivation was to take charge of IT and not to let M$ continue to lead Munich around by the nose. M$ forced Munich to migrate away from NT4. The city of Munich decided to what they were going to migrate. They chose GNU/Linux because it would allow them to decide when they would change OS and it would allow them to use open standards rather than M$’s lock-in file-formats.

  45. iLia says:

    I am sorry for my clumsy English but I have someting to say 🙂

    Ok, they want to migrate 14,000 computers to free and open source software, and they saved 4 million Euros, sounds good, but if we divide 4,000,000 by 14,000 we will get 285.7 Euros of economy per year or 23 euro of economy per month.

    And if we assume that the average salary of an office clerk in Germany is then the economy is only %1 in comparison with this salary, and the salary is not the only expense which an employer has, there is also rent which must be payed and other costs.

    And we should not forget that people don’t buy MS licenses every year, so the real economy should be even lower.

    It will be interesting to compare the economy of migration to linux with the annual turnover of a medium size company, but i am sure that this economy will be so tiny that it can be easily neglected.

Leave a Reply

Your email address will not be published. Required fields are marked *