Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - Paolo F Cantoni

Pages: [1] 2 3 ... 546
1
General Board / Re: ActiveX vs COMObject and Javascript
« on: August 04, 2022, 10:49:52 am »
PS. I hope they never stop supporting VBScript because that would mean years of code to be rewritten :-\
Unlikely to happen. What has happened is that the Microsoft components that EA relies on to debug VBScript and JScript are effectively dead. By contrast, the Javascript support is fully contained within our codebase, so that's recommended.

I use VBScript extensively (though not exclusively) because it has ByRef parameter passing.  I couldn't see how to do that in JScript.  Is it possible?
Yes, it is. All variables and arguments are assigned by value, but for objects, the value of the variable is a reference. So to pass something by reference, all you need to do is wrap it in an object and pass that.
(my emphasis)
I already knew that, but that was a shirtload of work to just get a string passed ByRef.  If I were a masochist, I could do the same in VBScript, so your comment is correct but useless.
Nevertheless, based on your point about the Microsoft components, we'll have to start planning the changeover.

Paolo

2
General Board / Re: ActiveX vs COMObject and Javascript
« on: August 04, 2022, 08:14:14 am »
Hi Martial,

Yes, try Sparx support and see what they say.

I do all of my scripting in VBScript, so I haven't encountered any of these issues.

Geert

PS. I hope they never stop supporting VBScript because that would mean years of code to be rewritten. :-\ 
Not wrong!

I use VBScript extensively (though not exclusively) because it has ByRef parameter passing.  I couldn't see how to do that in JScript.  Is it possible?

Paolo

3
General Board / Re: Evaluating EA 16: UML 2.x XMI export incredibly slow
« on: August 02, 2022, 08:16:33 am »
Hi Richard,
I believe Sparx used to steer people away from UML 2.1 transport between EA instances.  Even today, UML 1.x is preferred for XML transport.  UML 2.1, I suspect, is principally there for import from external applications.

I only use UML 1.x.  Increasingly I'm using the Native formats for inter-EA transport (which I understand you can't use in this case).

HTH,
Paolo

4
Hi Ian,


Are you using any automated testing at present?  In a couple of previous lives, I was a Testing Manager and employed automated testing to good effect.  Contact me privately and we can have a chat.


Paolo

5
For some of our metatypes, we want to enable the visibility of the Notes compartment when the item is created on the diagram (in Rectangular mode).
We tried using the Templates package to do this, but it doesn't seem to work.

Does anybody know how it can be done?  Via MDG?

TIA,
Paolo

6
As you'vedscovered, EA tends to "fail silently".
When this happens to me, I normally replace the link I specified with one of the links that work.  If that now shows the corrected icon, then it's likely an addressing problem with the original link.
Otherwise? ???

HTH.
Paolo

7
Suggestions and Requests / Re: Suppress Type icon in version 15.1
« on: July 28, 2022, 08:34:57 pm »
This has irritated me (and I suspect many) for a decade!  Anyway, yesterday during a "daydreaming moment" - when my mind was free to make arbitrary associations - I realised that they aren't "Stereotype" icons but Type icons.  No wonder they weren't being suppressed!
Hi Noel,

The UML 2.5 specification does NOT mention Type Icons, only Stereotype Icons.  Notwithstanding that, EA is self-inconsistent concerning the decorations it places in the top right of the rectangular notation.  So I don't believe they can fix this, given the current internal architecture.
The general solution adopted by most over the last decade is to obliterate the existing decoration with an appropriately sized and rendered rectangle and then create your own icon (or not as required) thereafter.

HTH,
Paolo



8

On our machine, we have both 32-bit and 64-bit EA installed.  When we try to access any .eapx repository with the 32-bit version, we get the following message when we try to open a diagram.
DAO.Database [0x00000d0e]
Invalid Memo, OLE, or Hyperlink Object in subquery 't_object.PDATA2'.

The diagram opens "Empty" - no diagram objects are visible.  Everything else seems to work OK.  As far as we can tell, there is NOTHING wrong with thet_object.PDATA2 column.  Since v16 64-bit didn't support MS Access, we didn't try.  However, in the last few days, Sparxian Eve mentioned using the  64-bit version of the Microsoft Access Database Engine 2010 Redistributable in another thread.  So we installed it and tried to access the same file.  Success!  No problem!

Does anyone else see anything like this?  It's pretty weird!

TIA,
Paolo

BTW: We can't install the 32-bit version of the Redistributable because we have 64-bit office products.

[Edit: Forgot to say that "Alles ist in Ordnung" with v15.2]

9
Paolo

Many thanks
Green is the colour that is expected if all working OK.  So looks fine.  In the rework to support 64-bit I think I've improved the checking that sets the colour.
Anybody find anything unusual/unexpected please let me know - email best - so I am made aware and I'll try to resolve.
Hi Adrian,
(I'll also send an email)
I have a separate AddIn (that coincidentally uses much of the same lower-level projects).  EAII v7 says all green for both, EAII v6 says Cyan for both, but one (the original) loads OK in EA-32 v15.2
the new one is Errored in EA with "0x80040154 - Missing".  I haven't worked through the problem on my end, but I thought you should know about the anomaly.

Paolo

10
We use the /?guid=<guid> feature EXTENSIVELY to target specific diagrams in our various HTML report structures (we have more than one).  "Punters" DON'T like having to navigate deep structures.  So we just send them to the specific diagrams we want them to consume.

We just exported our first v16 HTML report and the  /?guid=<guidvalue> is broken!

Please rectify this immediately!

Reported,
Paolo

11
Bugs and Issues / Re: Integrity V1605 does nothing for MSSQL
« on: July 21, 2022, 11:16:15 am »
Just to indicate how well thought out this change has been (that's irony, ); when you DON'T select any tests, the PIC reports "No problems found", NOT "NO checks selected"!

Sheesh!
Paolo

12
Having an object of the wrong type is not necessarily an integrity issue, so I'm not convinced that EA integrity check should necessarily correct it.  What would it change it to? There could be multiple options for the object type.

My thinking is that wrong object type is a design validation issue and I don't think EA has that feature.
Agreed that EA shouldn't necessarily correct it. 
But the feature is named "Project Integrity Check", NOT "Project Integrity Restorer". It has a restoration option, but that needs to be enabled by the user.  There may well be other Integrity issues that EA should detect, but can't rectify.  It should still list them and note that they cannot be rectified without external action.


Paolo

13
I received the following response from Sparx Support:

Generally speaking, the tests performed in the integrity dialog are related to ways that we know EA itself has failed in the past or anticipated likely points of failure.

Just because an Object Type value isn't known by the current version of EA doesn't mean it hasn't been added by a newer version that has also used the repository. In that case, trying to correct your issue would potentially be introducing new problems.

To which I replied...


This response was not unexpected.  "Generally speaking"  means some tests are more generic.  If its intent is principally EA-generated errors, then DON'T call it a "Project Integrity Check"!  Call it something like "Finding EA Stuff-Ups".
The question is how to solve the potential problem you mentioned?
As I understand it, we users are NOT allowed to add "unknown" object types (since EA generates the documented error).  The "known object types" should be listed in t_objecttypes.  If they are not, then that is a design defect, and the table's function is spurious!  I'll leave it up to Sparx to figure out which ones are missing (there are!).  I would suggest that comparing the list of object_types in t_object with the list in t_objecttypes, generally speaking  ;) , would be a good test to add - especially since Sparx have made this inconsistency!


What do users think?
Paolo

14
Report a bug. Won't help much but sending none will even help less :-/

q.
I need confirmation first.  I've had issues before which turned out to be our environment.

Paolo

15

When we try to change the password of a local user, we get an 0x80040E57 string or binary truncated error, and when we try to use the new password, it fails!  We have checked t_secuser (currently defined as:
Code: [Select]
CREATE TABLE [t_secuser]
(
   [UserID] nvarchar(40) NOT NULL,
   [UserLogin] nvarchar(255) NOT NULL,
   [FirstName] nvarchar(50) NOT NULL,
   [Surname] nvarchar(50) NOT NULL,
   [Department] nvarchar(50),
   [Password] nvarchar(12),
   CONSTRAINT [pk_secuser] PRIMARY KEY CLUSTERED ([UserID] ASC)
)
GO
) and the values DO change! Does anyone else get this?  Now that we changed the password, we can't get in as a local Admin!

BTW, even if we replace the password with the SAME password, the password value changes!  Should it?

Since (if you note), the password column can be NULL, if we set the value to NULL by direct SQL, we can log in as admin, <no password>!  Not very secure!  NOTE also that if we leave the password fields blank in the UI, we get a value in the Password column (and the error message) and can't log in as admin, <no password>!

If you're confused, so am I!
Paolo

[Edit: On a whim, we changed all the column lengths to much larger values but still get the error!  They are now back as they should be]

Pages: [1] 2 3 ... 546