Salient Soft

I was unable to access a couchDb from another computer on the local network.

There were 2 issues that needed resolving to address this:-

1/ By default, CouchDb binds to 127.0.0.1:5984, and therefore is only accessible if you use this address (i.e. the loopback adapter) explicitly. Other adapters, such as the lan adapter which has the local ip address, will not work, so you can’t enter the server machine’s own IP address as this won’t work. To fix this, edit local.ini in <couchdb home dir>/etc, and bind to 0.0.0.0 instead of 127.0.0.1, as  follows:-

[chttpd]
port = 5984
; Options for the MochiWeb HTTP server.
;server_options = [{backlog, 128}, {acceptor_pool_size, 16}]
; For more socket options, consult Erlang’s module ‘inet’ man page.
;socket_options = [{recbuf, 262144}, {sndbuf, 262144}, {nodelay, true}]
bind_address = 0.0.0.0

Note the use of the [chttpd] section in the ini file, not [httpd]

Editing local.ini will override settings in default.ini, and these settings will also be preserved across upgrades, whereas edits to default.ini will not.

This is described in this post here. Note that you can also edit configuration parameters in fauxton in the configuration section. Clicking on the value of a parameter will expose an edit box to allow changes.

Once changes have been made you need to bounce the service.

2/ My other issue was to do with Kaspersky firewall settings. To ensure that you can access a machine from elsewhere on the lan, make sure your network type for the lan adapter is set to trusted and not public. Public will force you to enter explicit rules for every address/application/port etc. that you use, which is a complete pain on a lan. Trusted just allows all machines on the lan to interoperate freely. This is detailed here under how to change the network status.

No Comments »

When adding a menu button to an ion-navbar, there is a styling issue whereby the menu toggle gets the wrong styling.
This depends on whether the menuToggle attribute is present.

If the attribute is present and the toggle is placed inside an ion-buttons group, it gets the wrong styling via a special case for menu toggles (flex rather than inline-block like the rest). This (incorrectly) puts it on a separate line above the others. Putting it separately above the ion-buttons group, with an explicit left on the button, fixes the problem.

If the menuToggle attribute is NOT present, then the button needs to go inside the ion-buttons group as otherwise for example for a left group the other left buttons go to its left. The rules for all this styling seem a bit quirky and not clearly defined.

This post discusses this issue.

No Comments »

I wanted to try showing a centered title with a back button on the left, in the NavBar.

I tried various ways of doing this with css hacks, using floated <h1> and <div> combinations, but could not get it all to line up correctly.

The correct way is just to include an <ion-title> in the NavBar, as follows. This can be centred via the text-center attribute.

<ion-header>
  <ion-navbar>
    <ion-title text-wrap>
      {{place.localisedName}}
    </ion-title>
  </ion-navbar>
</ion-header>

Whilst text-center worked correctly in the title, note that text-wrap did not and was ignored.

As I did not take this pattern any further (I decided not to use titles in the toolbars but to put titles immediately underneath) I did not investigate this further.

This highlights an important issue. Too much CSS hacking breaks the encapsulation of the CSS model used by Ionic/Angular which is a bad thing – other things are in danger of breaking even if it does work initially.

it is obviously better to go with the tools/tags provided and investigate this way thoroughly first for a solution – in this case I did not and initially went down a blind alley which took time as well as being a suboptimal approach which did not work and broke encapsulation.

No Comments »

This post is a common point to document basic errors of the above kind.

1/ Karma tests fail to run giving the following kind of error:-

‘places-list’ is not a known element:
1. If ‘places-list’ is an Angular component, then verify that it is part of this module.
2. If ‘places-list’ is a Web Component then add “CUSTOM_ELEMENTS_SCHEMA” to the ‘@NgModule.schemas’ of this component to suppress this message. (”
<div> Places results:- </div>

[ERROR ->]<places-list></places-list>

In this case the application ran but the tests failed as above, following the addition of a new component, PlacesList, which was added via ng new component, and used in the root AppComponent.

This StackOverflow post details the issue and the solution. In my case the new component needed adding to the declarations list in TestBed.configureTestingModule in app.component.spec.ts as the component was a required dependency. I had added it (or ng generate may have, I cannot recall) to the @NgModule declarations list in app.module.ts, which enabled the application to run. However the test bed needed it too.

2/ I hit a similar issue with the PlacesService. I was getting this error on the tests:

Chrome 56.0.2924 (Windows 7 0.0.0) PlacesListComponent should create FAILED
Error: No provider for PlacesService!
Error: DI Error

I used 2 different ways around this problem.

a/ I added PlacesService as a provider in  places-list.component.spec.ts – this enabled both the tests and the app to work.

beforeEach(async(() => {
  TestBed.configureTestingModule({
    declarations: [ PlacesListComponent ],
    providers: [PlacesService]
  })
  .compileComponents();
}));

b/ Alternatively, I could also add PlacesService as a provider in both app.component.ts (i.e. at the top level) and the test definition,  places-list-component.spec.ts

@Component({
  selector: 'app-root',
  templateUrl: './app.component.html',
  styleUrls: ['./app.component.css'],
  providers: [PlacesService]
})

beforeEach(async(() => {
  TestBed.configureTestingModule({
    declarations: [ PlacesListComponent ],
    providers: [PlacesService]
  })
  .compileComponents();
}));

Whilst b/ defines the provider in 2 places, from basic knowledge this feels better as in a/ the provider is defined in the component that uses the service which feels almost as closely coupled as if I was not using DI at all.

Interestingly, when I removed the PlacesService as a provider from app.components.ts and added it to the @NgModule as a provider, in app.module.ts, this failed similarly. This was strange as I expected app.module.ts to declare the provider globally (expect perhaps for the testbeds).

This needs further looking into and understanding of the DI concepts – on the face of it, the providers (which are the actual lists of components to be injected) should be as decoupled as possible from the classes where they are used. At the moment, a lot of what I am seeing looks far too incestuous, coming from a background in Java/Spring where Spring is entirely external and responsible for the ‘provider list’.

No Comments »

Further to getting the Tour of Heroes tutorial (which is systemjs based) debugging correctly in vs code (see here), I moved on to creating a new app via angular-cli as a basis for moving forward.

The point about angular-cli is that it uses webpack as its packaging tool, which is better than existing packagers for breaking down enterprise level applications. It mandates a standard for project file structure which was lacking in the past in angular, and creates everything according to that standard. It also allows additions of components etc. which will also conform to the same structure.

Previously I had settled on using vs code, as I had issues with all the previous IDEs I had tried (eclipse with angular 2 eclipse, webclipse) as per here).

I tried the following steps:-

1. create a new angular project via ng new
2. build/start it via npm start
3. run chrome in debug mode via the following shortcut "C:\Program Files (x86)\Google\Chrome\Application\chrome.exe" –remote-debugging-port=9222
4. debug in vs code using a suitable (clearly not as in the end I could not get it to work) launch configuration

The problem is that vs code could not map the .ts source to the .js. via the sourcemap, which is provided to do the mapping.
When adding a breakpoint, the following error occurs:-
“Breakpoint ignored because generated code not found”.

As vs code is widely used, there were a number of posts about this issue as follows :-

https://github.com/angular/angular-cli/issues/2453
https://github.com/angular/angular-cli/issues/1223
https://github.com/angular/angular-cli/issues/4526
https://github.com/angular/angular-cli/issues/2491

Unfortunately I could not resolve the problem via any of these.

A key point that these posts do not clarify on is that when starting with npm start or ng serve, any generated js file bundles (e.g. main.bundle.js) which are bundled js files under <project>/dist/) are deleted along with the <project>/dist directory, and the generated js is entirely served from memory. The IDE therefore has to get this js via a url, so all the discussion about tweaking the sourceMapPathOverrides to things like:-

"webpack:///*": "/*"
"webpack:///C:*":"C:/*" (for windows – substitute correct drive letter is the theory)

…seem to miss the point if they are aiming at target directories for the js. Of course, they may refer to the need to derive map keys for the sourcemap which would make sense if the sourcemap contains urls as values to return the generated js, but this is not made clear.
Either way, I could not get any of this to work in vs code and could not get a working breakpoint on the initial line of code that assigns the "app works" string.

In the end I dumped using vs code and gave Webstorm a try. This was immediately successful with almost no configuration needed – details as per this post.

No Comments »

The aim of this was to get the tutorial up and running in a suitable IDE, such that I could debug and step through the code whilst understanding it. The tutorial site is here. The standard repository on Git is here.

A number of different IDE options where tried, and most of them hit problems and were therefore discounted. The options tried were as follows:-

Eclipse with the Angular 2 Eclipse plugin

The following links document how to set this up:-

http://fcorti.com/2016/08/22/angular-2-eclipse-how-to/

http://stackoverflow.com/questions/35890887/how-to-get-angular2-to-work-in-eclipse-with-typescript

When loading the Tour of Heroes tutorial, I tried to import it as a file system project as the cloned Git repo was not an ecplipse project.

The problems I hit with this were as follows:-

1. After importing into eclipse, there were compile errors on import and export statements in rollup.js.
2. The core problem appeared to be eclipse bug 496348 which was listed as resolved but judging by the comments at the end was still a problem.
3. I tried various ways around this, such as upgrading to a dev release of eclipse Neon, trying a prerelease of eclipse Oxygen, and upgrading the eclipse plugins. None of these solved the problem.
4. As this was not a very commonly used IDE, there was not that much in the way of online help. In particular, no-one appeared to have posted online about getting the Tour of Heroes demo working with it. Therefore, support etc. going forward would likely be a problem.

Webclipse

1. This is a paid for eclipse plugin, also available as a separate IDE on its own called Angular IDE, detailed   here.
2. I tried this also with Eclipse Neon and Oxygen, and received similar Typscript related errors that I could not eliminate when trying to load the Tour of Heroes demo. I therefore dumped this too.
3. Whilst I did not try IntelliJ/Webstorm at this stage, which was the Jetbrains/IntelliJ paid for option, one Stack Overflow comment (which I now cannot find) cited that there were still Typescript issues with it, even in IntelliJ. (With hindsight, I wish I had, as in the end I have standardising on it as it works both for this tutorial, which uses systemjs, and for applications generated via angular-cli which use webpack. Details on this stage of the journey here and here).

Visual Studio Code

1. This is the IDE I settled on at this stage. Whilst it meant learning another IDE, it had a number of overriding benefits.
2. Basically it just works with no problems and no wacky Typescript errors! The basic point here is that Microsoft originated Typescript, and they are the authors of VS Code, so unsurprisingly it has the best compatibility.
3. As it is probably the most commonly used IDE for Angular, unlike with the other IDE solutions, there are many examples online for it, including for the Tour of Heroes tutorial, which needs a custom launch.js file for running the demo.

The setup steps for it were as follows :-

1. Follow the initial steps including installation of the latest Node.js and npm from the Angular tutorial site here
2. Rob Lourens has forked the Tour of Heroes git repo here, with a custom launch.js file for launching in VS code. Clone that git repo and follow his instructions.
3. Note that this uses the VS Code debugger for Chrome extension, in order to allow browser run angular JS code to be debugged inside VS Code instead of in the Browser.
4. In order to get this working, I had to start the app from the command line first via npm i (to build) then npm start.
5. Also, initially I could not get VS code to connect to Chrome, either by running up a new chrome via Rob’s launch.js configuration “Launch localhost with sourcemaps”, or by connecting to an existing chrome using “Attach with sourcemaps”. The issue was that Chrome needed starting with a command line option to allow debug connections via its debug port. This is detailed here (comment by auchenberg) – the command line is C:\Program Files (x86)\Google\Chrome\Application\chrome.exe –remote-debugging-port=9222
6. Once the above was done, I could connect from VS code provided I tried to attach to the existing chrome using  “Attach with sourcemaps” – otherwise, a new chrome was started which did not allow debug connections!
7. The steps needed were therefore 1/ run (having built) using npm-start, 2/ run up a debug chrome, then 3/ click on the debug icon in VS code, select the  “Attach with sourcemaps” dropdown launch.js option, then hit the green go button just to the left of it. This successfully fired up the demo.
8. Note that I was able to breakpoint in the code using VS code either on the Typscript (.ts) files or the equivalent transpiled .js files. It appears that the compiled code is interspersed in the same folders as the Typescript, which does not seem cool at all – something to look into. I would expect that the build artifact code would be separate as in Java.

No Comments »

Previously this had been an issue as the PC would sleep during the validation of a Backup, so I manually switched to an Always On power plan.

With the current PC build I was using, the PC would hang and not shut down cleanly if it was put into sleep and woken whilst an eSATA drive was attached to the 2-port STARTECH card I was using. Providing the PC did not sleep, all was fine.

I therefore looked for a means of automatically switching to always on for Backup runs. Rather than switch back to the normal power plan automatically after the backup, I elected to leave Always On set and switch back to normal on every startup. This way, I would be sure that if I still had the eSATA drive connected after a backup, the pc would not sleep.

This Acronis forum post which refers to this instructional pdf detail a workaround to the problem, which involves coding a batch file to switch power plans, and calling it as a pre/post custom command in Acronis in the backup settings.

In addition, as I wanted to set the plan back to normal at boot time, I used the technique from this StackOverflow post which worked correctly. Originally I had tried using a scheduled task triggered at boot time, but it refused to run and did not give a reason why.

My registry settings file, batch files, and vbs script are listed below for reference. Note that the GUIDS correspond to the particular power plans, and are unique for each one and different for each pc – see the above pdf for details of how to list the power plans with their corresponding GUIDs.

This fixed the problem, but I remain unimpressed that Acronis does not handle this automatically or have a setting for it in a backup definition. You can tell it to prevent sleep for scheduled backup runs, but not for manually initiated ones.

SetPowerPlanNormal.reg

Windows Registry Editor Version 5.00

[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Run]
"SetPowerPlanNormal"="wscript.exe \"C:\\System Management\\RunInvisibleBatchFile.vbs\" \"C:\\System Management\\SetPowerPlanNormal.bat\""

RunInvisibleBatchFile.vbs

CreateObject("Wscript.Shell").Run """" & WScript.Arguments(0) & """", 0, False

SetPowerPlanNormal.bat

:: Change Power Plan to Normal.
::
@ECHO OFF
POWERCFG -SETACTIVE ccaec46d-cbf8-42af-9e8f-ab66182942f7
::
@EXIT

SetPowerPlanAlwaysOn.bat

:: Change Power Plan to AlwaysOn.
::
@ECHO OFF
POWERCFG -SETACTIVE 499ab33e-0735-4605-8ccc-98211478164b
::
@EXIT

No Comments »

Update 15/5/2012

Further to the issues below, whilst I could run Sage satisfactorily for the most part, some features still needed ODBC to be working, such as Invoice preview and VAT return printing.

After a couple of attempts I was able to setup ODBC after the event by running E:\Salient Soft\Accounts\Sage\Instant Accounts\ODBC32\Disk1\Setup.exe. This allowed ODBC setup/install even after Sage had been installed.

Sage was happy even though there was no mention of the ODBC driver or connection for Sage under Administrative Tools/Data Sources (ODBC). The real underlying issue was not discovered, but the problem was finally resolved.

I was then able to successfully print a VAT return and preview invoices, which previously had failed.

Original Post

This was a reinstall using a fresh windows install, but the same location for Sage on a data disk. The following issues were hit:-

1. On install, it hung at the point where it said “Setup is configuring the ODBC”
2. I tried installing from a shortcut to the CD, with administrator mode set and Windows XP compatibility of various flavours, but the same error persisted.
3. I then tried removing the “ODBC files” option from the (custom) install. This changed the problem – the install now hung at the point where it said “configuring MIS files”.
4. I then did another custom install removing everything that I did not need, including ODBC, MIS support, E-Banking, web/email templates etc.
5. This time, the install finished successfully. Furthermore, Sage ran fine even though I had not told it to “configure the ODBC”.

Note that it is possible that the previous duff installs did configure ODBC and that the final install was dependant on the previous ones to work. This should be borne in mind if trying again. However, all the installed files for Sage were all there anyway as this was a reinstall – I was only doing it to install the windows/registry settings/shortcuts etc. that Sage needed.

I did not try a new install to a new directory as I did not need to – this may have worked, but I cannot say. If so, it may be possible to import or copy the old company data into a new install. A quick look at the menus and a quick google did not reveal any obvious way to do this, but a judicios copy may work.

Note also that this issue did not occur the first time I installed on Windows 7 64. This first time was also an install over an existing Sage installation copied from Windows XP.

No Comments »

I received this error whilst performing a commit via Eclipse:-

The server reported an error while performing the "cvs add" command. cvs server: xxx added independently by second party

The error related to some test database scripts in the project. I searched for solutions to this, and found some comments about tagging the files as merged, but this was not really any help.

In the end, I examined the files using TortoiseCVS, and noted that the offending files were not flagged as being under source control – they did not have any icon decoration on them.

In the end the following steps solved the problem:-

1. I made sure I had copies of them elsewhere, and then deleted them from the (TortoiseCVS managed) directory under the eclipse project.
2. I then did an Update with TortoiseCVS – this reloaded the latest versions from CVS, and this time they were correctly icon-decorated as managed.
3. I then copied the new versions in to replace these. When I did this, the new versions were correctly decorated in red as updated versions but known to CVS/Tortoise.
4. I then did a commit with Eclipse, and this now proceeded without error.
5. Finally to be safe, I did a file compare to check that the correct latest versions were in place.

This solved the problem. It appeared that somehow I had lost the required CVS metadata such that the files did not appear to be CVS managed. Tortoise was very helpful here – if you see files with no icon decoration, this is immediately suspicious.

No Comments »

Update 7/12/22

Previously I have implemented a composite iterator for iterating a tree of tags from a search on a PropertyTree implementation. Note that in this case, the abstract base class for the composites was a TagTree, and the concrete subclasses were a TagList and a Tag. As a Tag did not need a list, the list of children was in the concrete TagList and not the TagTree. Therefore, the TagTree did not need to have generic references to self referential subtypes in the way the example below does – in fact it did not need to be generic at all. Therefore it did not suffer this problem. For further clarity see the example below from Angelika Langer here. In her code, the list is in the abstract superclass which is why it suffers the problem below and potentially needs the infamous getThis trick per below.

Original Post

Some generic classes are recursively self-referential, such as in the following abstract class statement for a tree node :-

public abstract class Node <N extends Node<N>>  {
…
}

In this situation it is sometimes necessary for example to pass a reference to this from the superclass to a specific subclass which expects an argument of type N :-

public void methodA(N node) {
//do something with the passed node
}

public void methodB(N node) {
node.methodA(this); // error – incompatible types
}

In the above fragment, the error is because this in the superclass has a type of Node<N> which does not match N.

Another related use case is where a fluent API returns a ‘this’ reference from each of the fluent methods. If for example a fluent method in an abstract base class needs to return a ‘this’, as above the reference needs to be for the concrete subclass so the same problem occurs.

The topic is discussed in detail in Java Generics and Collections, section 9.4 on the Strategy pattern. The section from the book may be viewed online at Flylib here.

Angelika Langer also discusses it in more detail her Generics FAQ here

In the above case, the problem can be solved with the so-called getThis trick. An abstract getThis method is defined in the superclass, and implemented in each subclass:-

Superclass

protected abstract N getThis();

Subclass

protected Subtype getThis() {return this}; // where Subtype is the concrete subtype of N

Now the superclass can obtain a type correct reference to this and pass it without error:-

public void methodB(N node) {
node.methodA(getThis()); // this time it works
}

As Angelika Langer points out in her examples, there can be other ways around this kind of problem depending on the situation, but sometimes you do need the getThis trick.

No Comments »