sgolayfilt no longer support SINGLE
in [Matlab]
|
From: Jan Simon on 9 Jun 2010 17:27 Dear Bruno! > The Savitzky-Golay filter (Signal Processing Toolbox) no longer supports array of single type. > So my question is simply: who can explain "why"? > Bruno This does not answer your question. But here you can find a Savitzky-Golay filter, which is faster than Matlab's implementation and operates on SINGLEs: http://www.mathworks.com/matlabcentral/fileexchange/5661 Jan
From: sscnekro on 9 Jun 2010 18:47 Hi Bruno Luong, Jan Simon, Derek O'Connor and dpb, I am thankful to all of you for the occasional programming lessons you gave me on this newsgrop. In turn, I'd like to share a few remarks on this this thread from my point of view (as of an economist), as these are sorts of economic and legal matters. If at least one mathematician / technical engineer out there learnt something from this post, it would be worth the effort. I hit on the thread accidentaly. Either of you have been implementing Matlab in your commercial projects, involving all kinds of responsibility towards clients. You have experienced troubles with backward compatibility of the new releases, that could eventually have serious, far reaching consequences. Even despite difficulties, you have been defending Matlab against the team of your colleagues. Now, what happened? This thread, I mean. First, the inappropriately gentle message of Bruno posted to the newsgroup, instead of addressing them a letter or something they would *have to* deal with officially. Second, the phlegmatic *ignorant* replies of the MathWorks guy. Certainly they test each new release and certainly they do care for compatibility. It's rather the way they communicate the changes and especially, their notion on legal issues. This should be communication with a customer of Bruno Luong size on a matter as this one???! My impression from this happy company is that they do not know, or do not want to know, what *public relations* are about and in particular, I am afraid, they have no legal department at all, nor do they hire any legal services. Maybe it is not obvious to your eyes, or you are just not into it, but to my eyes as of an economist one thing is pretty obvious. If you'd get any of the difficulties you experienced to the U.S. court, you would make it to millions of dollars case!!! Just ask out any laywer to do it for $1 + 20% of the sum in question. They would stand queue for you. I hope I just dream and that this is not reality. Otherwise, did you continue then in this communication? How did it end? Please, Bruno, let us know. PS1 For Honglei Chen, a few tips: ------------------------------------------------------------ Dear Bruno, thanks for your communication. % describe briefly how you understand the issue raised % Explain the position of MW to the problem Should you wish to raise more questions on this problem, I will be glad to assist you directly at (...) \% e-mail, phone. Together with the team of my colleagues, we would appreciate the possibility to consult the difficulties you experienced regarding this and other parcitular issues with our latest release (...) ------------------------------------------------------------ PS2 For Bruno Luong - with service providers like MW try out this: *Am just f... upset!!* *Let me speak with your superintendent* *I'm really starting to be nervous* % List action you will take as consequence, let them feel they will loose you as a customer and could face legal process % Forget about polite and respectful communication for matters as this one. Otherwise they just continue in their winter sleep.
From: Jan Simon on 10 Jun 2010 06:52 Dear Sscnekro, > % Forget about polite and respectful communication for matters as this one. Otherwise they just continue in their winter sleep. The employees of TMW are human and they try their very best to keep Matlab stable, efficient and a source of money for them and for us. Some of their solutions are suboptimal under the user's point of view. Some changes in new versions are simply bugs, which are emerging in all complex systems. And some disadvantageous new "features" are caused by the limitation of the communication between users and TMW and between their software engineers and developpers. I do not see any reason to forget politeness and I do see a lot of reasons to intesify the respectful communication. I had a very polite conversation with my local Matlab distributor about the fact, that I strongly recommend several users of my programs to stay at Matlab 6.5.1 due to the limited compatibility and the amazing list of open bugs in Matlab 7. I've mentioned the need of a stable release, which is maintained by bugfixes for several years without introducing new features and new bugs. Then, and only then, I will suggest to buy new Matlab versions. The distributor can convert this information to $: 20 clients * (Matlab + Signal + DataBase + Statistics toolbox). Obviously this sum is not impressive and will not directly influence the strategies of TMW. I do not hear a lot of voices in this newsgroup calling for a stable release, but I assume there is a market for that. The distributor mentioned the zero-bug initiative in the cooperation with a large Japanese car manufacturer. I have respectfully explained my personal zero-buck initiative. Finallly some labs using my software had to update to Matlab 2009a for other reasons and I've ported the functions and started exhaustive tests. I had to care about 4 ugly incompatibilities (no VAX-D file support, bugs in END and BACKSLASH operator, rounding/truncation for DOUBLE->INT conversion), but I can use a lot of new features increasing the usability of Matlab and the programs. My conclusion: TMW is not in winter sleep. They have improved Matlab in the past and I hope they will do it in the future considering backward compatibilities. I'll politely repeat my appreciation of a stable release and I assume Bruno has good reasons for his friendly tone. Kind regards, Jan
From: sscnekro on 10 Jun 2010 08:07 Jan, what you wrote only re-confirms my point. May I go back to some of what you wrote and ask a few questions? Q1: Why is the communication btw users and TMW developers limited? Q2: Do you think, are the newsgroup posts a good representative sample of TMW customers viewpoints and needs? If not, can you give an example of such a communication channel between TMW and a broad customer base? Q3: In what ways has TMW supported you in the testing phase? Q4: Have you already succeeded in negotiating a particular change to TMW products? Why do you think the other customer was succesful in negotiating zero-bug? As a bottom line to this, I just hope, should you hear two guys talking on something in your field, with the first getting it wrong wrong wrong and the latter getting it wrong wrong wrong, would you not care, too, especially knowing how much it involves?
From: Jan Simon on 10 Jun 2010 10:48
Dear Sscnekro, > Jan, what you wrote only re-confirms my point. Fine. It was not my intention to disagree. Similar to your point I think that more communication between The Users and The MathWorks would assist improving the development. I'm really convinced, that politeness and respect are needed for a successful communication. The Dalai Lama said, that criticism and the formulation of negative emotions must be avoided as long as it does not lead to an improvement of its causes. Perhaps he did not focus upon the customer support of a programming suite. > Q1: Why is the communication btw users and TMW developers limited? In my opinion, the anonymous channels impede the communication, because they are not the biological way humen are used to talk. Sending a request to support(a)mathworks or files(a)mathworks is ever a kind of hard-bitten. But if I mail e.g. Helen or Shari directly, the answers are quick, direct and helpful (thanks again!). I've sent some ideas for vectorizing FILTFILT and a 65% faster version of IND2RGB to support(a)mathworks. The polite guy forwarded the message to the "developper board" and there it waits and wastes diskspace for the last 18 month. But whenever I had a direct and personal contact to a developper or another employee of MathWorks, they have been ever interested, quick and helpful. The distributor I've mentioned in the last post called me at Friday 6 p.m. to give me new informations. > Q2: Do you think, are the newsgroup posts a good representative sample of TMW customers viewpoints and needs? If not, can you give an example of such a communication channel between TMW and a broad customer base? I do not know such a channel. ImageAnalyst spoke of a meeting with the group for image processing. There are some meetings between the "Japanese car manufacturer" and TMW (I assume this deals with more than 20 clients * 4 toolboxes = x$). I think, the contents of this newsgroup is not representative, e.g. the "0:0.1:3 == 0.3?!" questions. But some posts caught the attention of TMW in the past. > Q3: In what ways has TMW supported you in the testing phase? TMW delivered the free test versions of Matlab and the needed toolboxes (some more than I checked out on the Web page). After some posts in this newsgroup they answered, that the VaxD support has really vanished without any notes in the documentation. The list of open bugs is really helpful for me. I've been very surprised that TMW did not accept a bugreport concerning ANCESTOR without knowing my license number. > Q4: Have you already succeeded in negotiating a particular change to TMW products? Why do you think the other customer was succesful in negotiating zero-bug? I do not exactly know, what the zero-bug initiative is. As far as I understand Kurt Goedel it is impossible to create a powerful program with zero bugs, because there is always an input which let the algorithm explode. Some bugs of 2009b are fixed in 2010a, e.g. described in: http://www.mathworks.com/matlabcentral/newsreader/view_thread/264225 Nevertheless, it is not an option to upgrade again, because the testing takes about 2 month and 2010a contains new bugs, and 2010b, and 2011a, and 2011b, ... A long-term-support version would be a great benefit and I'd expect that the experiences growing from such a version would help to improve the whole product. What specific problems did you have with which tools? What are your experiences with the TMW support team? Kind regards, Jan |