Bad design

I might be way off the mark here, but bear with me a second...
Photoshop CS5, Camera Raw 6.1, and now Lightroom 3 can all use Lens Profiles to correct images. You can share profiles with the Photoshop community is if you have Lens Profile Creator, but you can only download them if you have Photoshop CS5. Is this correct?
Presumably then, people with Lightroom are unable to download new profiles, and are unable to access the Adobe profile server. So what do they do? Share profiles over the internet in an ad hoc way? Isn't Adobe missing a trick here? Surely there are an increasing number of Lightroom users who are not sharing their new profiles - after all, what's in it for them?
If there are plans to add lens profile downloading to Lightroom, then fair enough. It's a shame it wasn't implemented earlier. But that's just tought luck, I suppose. I only mention this because I keep seeing messages of people requesting profiles by email or yousendit. It seems a terribly wasted opportunity for everyone else.

Well, I don't use Lightroom, and I can get CR lens profiles through Photoshop CS5, so it's not something which affects me directly. It does affect me indirectly though, as there seems to be an unofficial trade in lens profiles by Lightroom users who dont use CS5. So the altruistic profile exchange system set up by Adobe as an asset for their products isn't working to its best.
However, if the developers were limited by schedules, then this perfectly explains the situation, and I wouldn't expect many to have to make a feature request to make it obvious that Lightroom users need this feature too, for theirs and CS5 users benefit alike.

Similar Messages

I think I found a bug/bad design - what am i missing?

Hi Edwin and the team,
This is my first post as I'm sinking my teeth into the PM Server and Designer. We're considering to use the BPEL technology in the BIG way. And I think it's very cool as the Web Services-based EAI delivers a cost-effective and not proprietary solution!
Let me describe to you what I believe is a bug or a dubious design. In the Designer, File>New>Oracle BPEL Project creates a new project with a file bpel.xml. For example, look at your CreditFlow sample. In that file there is <partnerLinkBinding name="..." <property name=wsdlLocation"> http://<host name of the WSDL host server>/....
when the partner WSDL is hosted on the localhost that <host name> in the line above is the name of my machine instead of 'localhost'. Then, <project name>.bpel validation crashes when you're in the off-line (disconnected) mode or connected via VPN gateway (since obviously you think you don't need a proxy because everything is local). But, even if it works inside the LAN I think it's a bad design; since it's generated automatically by Designer the developer doesn't have a clue what's wrog when he/she needs to debug your code. The Designer should stick there 'localhost' instead of the the name of my computer on the network.
Here how I spotted it. I generated you CreditFlow sample in the Quick Start Tutorial. I went by the book, Compiled the CreditFlow.bpel and it crashed! The book did not say it was supposed to crash. The book said everything should be honky dory. The Validator printed a message: "connection timeout...." Since I'm new to the BPEL it took me awhile to figure out and once I stick 'localhost' in the right place in bpel.xml everything worked. Again, I was running everything on localhost.
If I helped you guys - great or what am I missing here?
The problem is that if my company decides to convert the Biz Analysts into BPEL modelers how could a BA figure out the solution to the problem like this? He/She would just give up and tell the Technologists (us) that this is a bad product.
Thanks,
Greg
P.S. What happened to Doron Sherman (CTO) and the old Collaxa folks? He is not listed in the ORCL directory. For those who don't know BPEL Server and Designer were developed by Collaxa and ORCL bought that company last year.

Hi Gregory,
I think that the problem you are reporting is a proxy configuration error. There are 2 places where you need to configure the proxy: in the obsetenv.bat and in the eclipse>windows>preferences>BPEL Designer.
If the server is up and the proxy configuration is correct, you should not receive a timeout error.
Let me try to explain why it is not possible/desirable to have localhost: the url is generated by the server when it generates a WSIL list of the processes deployed on the server. The server does not know if the client requesting this list is local or remote, this is why he uses the real host name (otherwise remote client would not be able to connect to the service).
One additional note: the tool is not currently targeted at business analyst (unless they have some development background). BPEL entails some inherent complexity: parallel processing, async interactions, exception management, WSDL, XML Schema, transformation. So although we are working hard on continuously simplifying the implementation and increasing the productivity, we are still targeting application developers.
I hope this helps.
Edwin

CISCO 2125 bad design by Cisco, product development should revise

I noticed Cisco product sales support staff is pushing 2125 WLAN controller as great WLAN solution. They got me into considering one as something what will work just fine.
I strongly disagree, this product is extremely poor design and it is missing gigabit port for LAN connection.
This product will support 25 APs with its eight 10/100 Mbps ports. Only 6 APs can be linked per port. So for 25 APs you are already using 5 ports. Per manual POE ports 7 & 8 should not be used for anything other than single AP so that leaves you with only 1 port #6 to connect to your LAN.
Yes, that is true, 100Mbps for all the WLAN users on the 25 access points and product is advertised to support high speed N wi-fi capable of doing 300Mbps per user node. What a nonsense? makes no sense even for lightweight use.
Makes you wonder who approved such product for production and what a bad design on the $6K piece of equipment
CISCO please put a Gigabit port on the 2100 series or stop marketing and pushing such poorly designed product.

“Thank you for your question. This community is for Cisco Small Business products and your question is in reference to a Cisco Elite/Classic product. Please post your question in the Cisco NetPro forums located here: http://forums.cisco.com/eforum/servlet/NetProf?page=main This forum has subject matter experts on Cisco Elite/Classic products that may be able to answer your question.”

I buy an iphone 5 bumper for my black iphone to protect it and now ive checked it after a few weeks it has infact scrated the black metal ??? bad design or cheeply made?????

bad design what is up with all this ?
it has inpact caused more damage than if i was to leave loose in my pocket

The bumper does not protect the back casing. If you wanted to protect that portion of the iPhone, you should have purchased a different type of protective case.

It's working, but is it a good or bad design?

hiya
I hope I�m not breaking any forum rules, but any help would be greatly appreciated.
I haven�t really done any programming so far, so finally writing an app with more than few lines of code was something different. Anyways, I managed to write it, but truth be told, I have no idea whether my design choices ( I�m not sure if that�s the right term ) were somewhat good or completely off. Thus, if I don�t know what parts of program are designed badly ( and why they are considered bad), then I won�t know where to improve. So I�m hoping someone could point out ( at least the most obvious )design mistakes I made.
It�s a simple �chat� server where several clients ( up to twenty ) can connect to the server and talk to each other. In essence:
1.server creates new client thread for each newly accepted connection
2.if one of these client threads receives some data from the client, then:
3. this client thread creates new ControlThread object (and passes received data to it )
4. ControlThread object in turn spawns a new thread
5. Finally, this spawned thread will send data to all client apps.
I must point out that I could handle some things better using collection classes. But ignoring that, I�m most interested in other design flaws I made. For example:
a) how scalable is this server app?
b) what would you�ve done differently ( in terms of basic program structure )? Perhaps creating ControlThread object each time data needs to be sent to other client apps is a bad idea
c) etc
package serverchat;
import java.net.*;
import java.io.*;
public class Main {
    public static void main(String[] args) {
        Server.runAll();
class Server{
    static int servPort = 1600;
    static ServerSocket servSock;
    static Thread controlThread;
    static ClientHandling[] clientClass=new ClientHandling[20];
    static int numClients = 0;
    static Socket clientSockets[] = new Socket[20];
    static void runAll(){
        initServer();
        handleClient();
    static void initServer(){
        try{
         servSock = new ServerSocket( servPort );
         catch(Exception e){}
    static void handleClient(){
        try{
            while(true){
              clientSockets[numClients] = servSock.accept();
              clientClass[numClients] = new ClientHandling(clientSockets[numClients]);
              numClients ++;
        catch(Exception e){
/*Each time some clientHandling thread receives data from client,
it creates new ControlThread thread. This thread then sends this data
to all clients connected to this server*/
class ControlThread implements Runnable{
    ClientHandling[] clientClass=new ClientHandling[20];
    int numClients;
    String message;
    Thread t;
    ControlThread(String message){
        this.message = message;
        this.numClients = Server.numClients;
        this.clientClass = Server.clientClass;
        t = new Thread(this);
        t.start();
    public void run(){
        System.out.println("ControlThread has started");
               for(int i= 0; i < numClients; i++){
                   if( clientClass[i] !=null ){
                       tellClients(i);
    /* sends data to all the clients. It does this by calling contrToClient()
     on an object that received this data from the client*/
    void tellClients(int i){
           clientClass.contrToClient(message);
class ClientHandling implements Runnable{
Socket clientSocket;
Thread t;
String message;
OutputStreamWriter clientWrite;
InputStreamReader clientRead ;
BufferedReader readBuf;
BufferedWriter writeBuf;
ClientHandling(Socket s){
clientSocket = s;
startThread();
void startThread(){
t = new Thread(this);
t.start();
public void run(){
try{
clientRead = new InputStreamReader (
clientSocket.getInputStream(), "utf-8");
readBuf = new BufferedReader (clientRead);
catch(Exception e){
do{
try{
clientSocket.setSoTimeout(1000);
message = "";
message = readBuf.readLine();
if ( !message.equals("") ){
informContrThread();
System.out.println(message);
catch(Exception e){
}while ( !message.equals("Stop") );
try{
clientSocket.close();
catch(Exception e){
/*this is the actual method that gets called by ControlThread object and
sends received data to all the client apps*/
void contrToClient( String message){
try{
OutputStreamWriter clientWrite = new OutputStreamWriter (
clientSocket.getOutputStream(), "utf-8");
writeBuf = new BufferedWriter (clientWrite);
catch(Exception e){
try{
writeBuf.write(message + System.getProperty("line.separator"));
writeBuf.flush();
catch(Exception e){
try{
clientSocket.close(); /* I assume I should close the connection
with client if write() throws an exception? */
catch(Exception e1){}
/*creates new ControlThread object, which in turn will
make sure all client apps receive this data*/
void informContrThread(){
System.out.println("informing contrThread");
new ControlThread(message);
thank you

1.server creates new client thread for each newly accepted connection Good. Make sure the server doesn't do any I/O in the accepting thread, including even construction of input and output streams.
2.if one of these client threads receives some data from the client, then:
3. this client thread creates new ControlThread object (and passes received data to it )Why? What does the Control thread do that the client thread can't do? Does the client thread really create a new Control thread per piece of received data? What's the purpose of the Control thread here?
4. ControlThread object in turn spawns a new thread Again why?
From your description you appear to be creating:
(a) a thread per connection
(b) a thread per piece of received data
(c) another thread per (b)
What you probably need for a chat server is an input thread and an output thread per connection. The output thread should read a queue and the input thread should write to the appropriate queue(s), making sure you don't create a loop by sending the client's own data back to him.
5. Finally, this spawned thread will send data to all client apps.No. Use one output thread per connection, as described above. As writing to a socket can block, you shouldn't use a single thread for writes to more than one client, otherwise the whole system can stall due to one non-co-operating client.
a) how scalable is this server app?It's about as non-scalable as it could possibly be. You will have an explosion of threads per item of data received if you've described it correctly, and you already have too many threads per connection.
b) what would you�ve done differently ( in terms of basic program structure )?Almost everything: see above.
Perhaps creating ControlThread object each time data needs to be sent to other client apps is a bad ideaMost definitely.

Components inside components - Bad design?

Is it bad practice/design to nest components inside components using Flex 4? Should I simply be creating components and inserting them into my main application as below, or doesn't it matter?
<com:MyComp1>
    <com:MyComp2>        <com:MyComp3>
            <s:Label text="This is a test."/>        </com:MyComp3>
    </com:MyComp2></com:MyComp1>

It all depends. Knowing nothing about the components, all I can say is that it depends on the complexity of the components individually and collectively as well as what level of reuse you may desire.

God or bad design?

I often inject an implementation of an interface into another class's constructor as shown below:
public void start() {
    Connector con = new DataBaseConnector();
    con.execute();
    ChartGenerator generator = new ChartGenerator(con);
    generator.process();
}the method 'execute()' on the DataBaseConnector class basically just setup different connections and create data in some data structures. This prepared data can then be used in the ChartGenerator class which calls various getters on the 'con' instance.
I don't use the Connector after its passed to the ChartGenerator.
I use this kind of design many places in my code to reduce the size and responsibility of my classes but I have the feeling that its not very Object Oriented (basically its just a way to wrap and pass data down the execution path).
It also seems a bit clumsy to remember to call the 'execute()' on the implementation to be injected so I am currently moving the steps to be performed to the constructor of the class to be injected. This way the data is always guranteed to be ready when the class is created using new.
But is the above 'pattern' a good or bad approach and do you guys use another apporach?

Good :o)
I use this kind of design many places in my code to reduce the size and responsibility of my classes but I have the feeling that its not very Object OrientedI don't understand your feeling.
I have the feeling that its not very Object OrientedThat has no importance whatsoever. What matters is whether it works, and whether it is easy to maintain (order may vary if you ask several people). OO is generally a way to make programs more maintainable when used properly (and less when improperly).
It also seems a bit clumsy to remember to call the 'execute()' on the implementation to be injected so I am currently moving the steps to be performed to the constructor of the class to be injected. This way the data is always guranteed to be ready when the class is created using new.That's usually more maintainable to have classes whose instances are properly usable as soon as the constructor returns. So yes it's a probably a good move. But that doesn't mean the former design was inherently wrong.
AFAIC the execute() method (not a very good name, it seems) is implementation-specific; so it appears in the DataBaseConnector interface and not in the more general Connector interface. So at least it's a good thing that the main class calls it and not client class (+ChartGenerator+): after all the main class (from the look of it I like to call that an Assembler or a Composer class) is the one that knows which implementation has been used.
But yes, pushing it further, the implementation-specific detail could be managed by the implementation class's constructor itself.

Monitoring badly designed DML job

Hi dear community
we have these dml job that basically it is something like this
set echo on
set serveroutput on
declare
   dr_range number := 9; -- for real run some_big_number := 9999
begin
for i in 0..dr_range
loop
    dbms_output.put_line('i='||i);
    slow_proc(i);
end loop;
end;when we did the unit test we only test the range in a very small window maybe 1 or 2. It works
However, when we runs it a large interval. It is very slow. I made an nohup overnight job and the job dies when I come back in the morning.
And another co-worker (the guy who design the procedure observe the same)
I concerned the dml job can have locking and commit issue that cause that. Also on top of that the dev env data is badly set up.
I would like to do a sql_trace with level-12 and see whats going on
or use plsql tracing ?
Or perhaps use ash ?
what would you guys recommend ?

vxwo0owxv wrote:
we have these dml job that basically it is something like this
declare
dr_range number := 9; -- for real run some_big_number := 9999
begin
for i in 0..dr_range
loop
dbms_output.put_line('i='||i);
slow_proc(i);
end loop;
end;
There is a fundamental problem with this code - DBMS_OUTPUT.
DBMS_OUTPUT has a static array that put_line() adds lines of text to. The more DBMS_OUTPUT is called to "write" debug text, the bigger this array grows. Until such time that the current server execution completes, and control returns to the client, allowing the client to read, empty and render this DBMS_OUTPUT array.
The array resides in PGA. This means inside the stack space of the server process executing the code. Causing private memory of that process to grow - and grow.
So what the above approach does is, for many iterations, call DBMS_OUTPUT and grow this array. Typically, your slow_proc() will also make fair use of DBMS_OUTPUT. Which can mean a large chunk of text written into the DBMS_OUTPUT array per loop iteration.
I have seen such an approach, ran as a DBMS_JOB, crash a RAC server node due to consuming all available memory and then some more. Causing such excessive swapping, that even the kernel struggles to get a CPU slice.
So perhaps that slow_proc() in the above template is slow. But do not disc0unt the impact on server performance caused by extensive use of DBMS_OUTPUT that results in a significant increase in page swapping.

Is polymorphism a bad design for object oriented concept?

i think polymorphism is bad concept for object oriented concept..having the same name with different parameter function ..according to my opinion may constitue in efficient skill in programming..instead if using the same name why we cannot use different name...for example each human being will have have unique characterstics..(i.e each persons DNA would be different..)so why can't we have different function names.. even if u choose a username to create a new account it would not support same names..i think it is confusing .. i think this concept is lessly used in modern programming usage... as it has less usage i think the future language designs should not support polymorphism..
i just want to know ur opnions..

Is it just me or did Sun fuck up the posting screens?I didn't notice anything; care to enlighten me?800-pixels-wide window and still a vertical scrollbar, but, to make up for
that, 250 pixels of unused whitespace to the left. And the search text
field is still unnamed.I agree with that rediculously wide left margin, i.e. rearranging some of
those things would allow for a better screen estate usage. But didn't Sun
already give us that crap in a previous incarnation of their mannah?
Besides that, I still don't notice any difference ...
kind regards,
Jos (<--- call me a silly ignoramus)

The Location bar shows me the TITLE tag info for a page in large font + the URL in small font. How can I turn off the display of the Title tag, I just want to see a list of URL's. Bad design to force it that way, I don't like it. I know what a URL is.

'''YAHOO'''
http://www.yahoo.com
I don't need the YAHOO. I find it so completely annoying that you have forced me to look at TITLE tags when I am trying to find the URL I want. This ought to be a setting that I could control. Is there a hack into my registry, or something along those lines, so I can turn off this horrible design element.

Mozilla hasn't changed anything like that in Firefox 3.6.x versions, which is getting security patches but not any further development of new or existing features. Plus, Mozilla doesn't install anything related to Yahoo, something else probably installed that.
Please provide a screenshot of what you are talking about. <br />
https://support.mozilla.com/en-US/kb/Adding+screenshots <br />
There was a feature similar to what you are describing, that was in and out of the Firefox 4.0 betas last fall, but the 3.6 versions never had anything remotely like what your description has me visualizing.
Ate you sure that one of your add-ons isn't doing that? <br />
http://support.mozilla.com/en-US/kb/troubleshooting+extensions+and+themes

A Finder bug (or bad design) -- cmd+del while renaming

This reproducible behavior is quite confusing:
1. In a Finder window, click on a file twice to enter the renaming mode.
2. Type some new file name.
3. While still in renaming mode, press Command+Delete.
4. The file will be moved to the Trash, yet in the Edit menu it says "Undo Rename".
5. If you choose to undo, the file will be un-renamed, but stays in the Trash.
What I wanted to do in step 3 was to press Fn+Delete, in order to produce a backspace, and when I accidentally pressed the wrong buttons, the file disappears, and when I use Undo, the file doesn't come back. Worst yet, when I clicked the Trash icon in the Dock, the Trash window doesn't come up, but that's another bug (which killall Dock doesn't help).
The above experience was very confusing. I suggest disabling the Cmd+Delete function while in renaming mode. Also, multi-step undo would be nice. Visual cue/animation for all file operations would be great too.
Ever since I upgraded my Tiger to Leopard, there are many bad experiences every week that makes me want to kill Leopard (if it's some hurtable thing..). I used to love Mac OS X. Now I am considering switching back to Linux.

Of course whenever we get confused by something we think that the interface should be changed to address that specific issue but every possible user confusion can't lead to an amendment of the system.
Specifically:
Multi-level undo means that we wouldn't have redo
The described behaviour only happens when you have made changes to the name before Command+Delete and that means you are suggesting that the interface would behave differently when a file name is changed until you have clicked something else to get out of that mode (which could be a problem for the way I work)
Although there is no animation, there is a distinctive click when something is transferred into Trash
My vote goes for the current behaviour.

Good or Bad Design

The following is the design for a reporting module. Any suggestions on the same would be appreciated.
1) ChartGenerator implements Generator.
Attributes �
private int ncompID = UNDEF
     private int njobId = UNDEF
     private String storedproc = ""
     private java.util.Date fromdt = null
     private java.util.Date todt = null
Methods �
generateTrendlinevalues(String[] entities2plot)
generatePieChart(String[] entities2plot)
generateBarChart(String[] entities2plot)
generateBarChart(ReportingLabelsAndValues labelvals,
String[] boards)
2) GenerateValues implements Generator.
Attributes �
     private int ncompID = UNDEF;
     private int njobId = UNDEF;
     private Date fromdt = null;
     private Date todt = null;
     private double[] values = null;
     private String[] labels = null;
Methods �
public BaseObjectList generateData(String storedproc)
public String[] getLabels()
public double[] getValues()
3) ReportingSource extends baseobject
Attributes �
String label
Double value1
Methods �
getSetRepSource()
4) ReportingSourceDAO extends baseobjectDAO
5) Servlet � responsible for generating the chart
Code in the servlet
GenerateValues gvalues = new GenerateValues(compid, jobId, frmdt,todt);
ChartGenerator obj = new ChartGenerator(compid, jobId, frmdt, todt, storprocname);
String[] boards = req.getParameterValues("boards");
ReportingLabelsAndValues labelvals = obj.generateTrendlinevalues(boards);
XYChart cxy = obj.generateXYChart(labelvals, boards);
PieChart cpie = obj.generatePieChart(boards);
XYChart cbarchart = obj.generateBarChart(labelvals, boards);

The following is the design for a reporting module.
Any suggestions on the same would be appreciated.To create reports?
There are certainly commercial libraries that do this and I believe there is at least one open source one as well, so the first suggestion would be to use one of those instead of creating one.

Tuning a bad designed application

Hi all,
i have to speed up a procedure that runs on a single serverver process (single connection).
The hardware has 8 cores, but the app runs only on une core. (as i supposed)
Is possible to split the process on multiples cores?
any idea, any trick?
cheers
Marco

> I am looking a way to increase the parallelism, without changing the code.
That is like trying to build a bird that flies and not being able to design wings.
There is no magic wand to turn code into parallel enabled and capable code. In fact, when using massive parallel processing (MPP), one has to design and code specifically for MPP. And this is very different from the normal design and coding.
So how does Oracle do it? Oracle does not turn client code into parallel code. Oracle can service certain client requests in parallel. If the client request to read the full table (e.g. select * from emp), Oracle can decide that given the volume of data to read, to perform the read in parallel.
And that is it. Oracle's parallel feature is simply doing I/O in parallel when the client request a large volume of data and the existing conditions allow that to be done in parallel. (conditions include sufficient idle PQ processes, etc)
What type of client requests can result in parallel I/O? Typically full table scans and full index scans.(and these are usually exceptions in the common application)
Many developers seems to have a very incorrect view of how parallel processing works in Oracle - seeing it as some dark magic or special performance juice that can simply be turned on.
Nothing can be further from the truth.
I suggest you refer to the [url http://download.oracle.com/docs/cd/B19306_01/server.102/b14220/bus_intl.htm#sthref2488]Oracle® Database Concepts's Overview of Parallel Execution.

6131 is power hungry, bad design?

I am disappointed by the battery of the 6131. It is worse than I could ever imagine when I read the reviews mentioning the "battery" problem.
It can't even last 12 hours. I don't spend the day talking...I check my mail a couple of time a day, talk a little, less than 30 minutes per day. It even drains overnight when the phone is off. What is wrong with this phone?
Any suggestion on how to reduce the consumption? Also, does anyone know if there is a design problem for that phone? Some kind of power leak? Anyone had their phone replaced?
The phone is great, the battery is horrible.

09-Sep-200704:11 PM
nj15 wrote:
08-Sep-200709:24 PM
korngear wrote:
Right,infact they should start it immediately.
And 1500 MHz should be the atleast.
The more the capacity, it means bigger battery size.Lets take Nokia batteries as an example.They should start making more phones with a Li-Po..regardless of milliampere count.
How I wish Nokia can make a slim phone with a BP-6M battery!!!Message Edited by nj15 on 09-Sep-200706:12 PM
Message Edited by nj15 on 09-Sep-200706:14 PM
Is(not) it possible to make sizeable battery with more power?

Is lightning cable a bad design?

I have gone through 5 cables within my household. (2 Iphone 5's). 2 were original and 3 aftermarket. There does not seem to be a difference between durability, if anything the aftermarket with braided cover seemed to perform the best. This is new to the Iphone 5, the cables for previsous Iphone/Ipod products lasted until the product was replaced. (we acturally gave away cables to friends as spares). So my questions is what is Apple going to do about this? I am pretty sure that Steve Jobs would not have sat on this and not taken swift action to permanently fix this problem.
Is there a fix? Is the design of being able to install the plug in either side the issue that makes it so delicate? Does this mean that we will have to wait for an Iphone 6 to resolve?
I am very dissatisfied with my first Iphone.

Alan, I did take the two cables back, they were quite well prepared with the replacement cables and admitted that they had quite a few of the replacements for the new cable whereas the older style not so much. They were not immediately prepared to replace the cable that had simple broken off inside the phone but were happy to replace the one that stopped working. Upon pushing for a replacement for both they agreed.
So for anyone with a lightning cable that either stops working or works intermittently, Apple will replace them no hassle. For anyone who finds the idea of replacing these cables regualarly, to be a problem, no solution has been found.

Re: Power Supply Problems/ Bad Design

The moderator of this site seems to think that any external power supply can have the same problems as apple powerbook supply. "Not so", say I ! Most PC notebooks and yes, I do use them, have a fairly sturdy cord on both sides of the connection. Should one find oneself in a situation without a power supply for a pc notebook it is usually possible to get one from a local Radio Shack or international electrical store. Not so with apple powerbooks. the connections are specifically apple and although there ae a few generic replacement PSU they are not generally available only from mac speciality stores. Why apple decides to design what is obviously a flimsy (my electrical gaffer on my movie looked at it and just laughed)and possibly dangerous device is beyond me. It occurs to me that with all the innovative design apple has come up with in the past has gone to the designer's heads. Now we are faced with a 'magnetic' connector on the new powerbook pro line. Are you kidding apple ? Please just give us a simple, generic power supply and conmnector so that we travellers can rest in peace knowing we can replace the PSU anywhere in the world.
BTW my PSU was smoking and sparking when I awoke in my hotel room until it just gave up.
PBG4 1.25MHZ Mac OS X (10.3.7)

Two alternative power adapter manufacturers:
http://www.lindelectronics.com/cgi-bin/store/shop.cgi/!ORDERID!/ACAdapters/Apple/dbx_gen_mfg_ACAdapters?industry=
http://eshop.macsales.com/item/Newer%20Technology/APLN2M/
The most common issues people have with Apple's adapters that I've seen are (not that this applies to you, but perhaps one will refer to something you've done that you didn't realize could cause harm):
1. They pull by the cord and not the bulb.
2. They push in without looking.
3. They let the cord bend at the bulb or at the brick. Give it at least two inches of slack.
4. They don't remove the plastic wrapping that is clear from the brick when they first get it. This under the wrong conditions could create an overheating condition.
5. They rely on unreliable power sources. Some older houses have issues with giving reliable power to outlets. Using a backup battery surge protector like those from APC can help.
By the way, my father's Dell power adapter had to be replaced too. Non-Apple notebook power adapters are actually more flimsy than you give them credit for. Part of the problem is designing the transformer to be in such a compressed space that is lightweight too. Until people don't mind the extra weight of a full fledged desktop power supply inside a notebook, thus allowing standard power cables, or a room temperature superconductor is invented, this will be a persistent issue. Take good care of your cords, and don't let them bend so much that they fray.
I might add, I went to Radio Shack, and they had the parts for me to build a power adapter for my Powerbook. Unfortunately the transformer brick would have been 3 times as large and as heavy.

Bad design

Similar Messages

Maybe you are looking for