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 - timoc

Pages: 1 [2] 3 4 ... 11
PCS General Board / Re: Host WebEA on a linux Server
« on: August 09, 2020, 10:45:29 pm »
Could not I just put the WebEA folder on my linux apache root directory and change the models settings in webea_config.ini to point to the PCS server ?
No idea.

PCS General Board / Re: Host WebEA on a linux Server
« on: July 28, 2020, 09:21:06 pm »
Hi guys,
is it possible to host WebEA on a linux server ?
No idea. I know you can run the RAS server under wine with some tweaking.

Keeping PCS Service on a windows server and link it to a linux Apache / PHP  ?

Did any of you tried this configuration ?
i know you can setup an nginx as a web facing proxy to another service. I have never tried it with any Sparxian stuff.

The MDG has not been updated since 2018, either the vendor thinks it is perfect (and it clearly is not) or it is a marketing trick to tick a box.

I have always thought that Sparx could vastly improved their import/export capabilities.

If I find time I can try to find out the profiles I create when I looked into this and see if attributes and complex hierarchies can be imported.
I too have noticed it is more trouble to use than other similar extensions i have tried. I originally had hopes of mining existing documentation to build a repository, also possibly having a round trip workflow with design documents. I have not yet been able to do anything usefull with it yet. Has anyone had a experience with excel and word? can they share?

As it is one of the two features that make the difference between Corprate and Ultimate edition tier payments, it should be more than a checkbox. I should have thought it would be able to do more, like supporting Office365 online by now...

Bugs and Issues / Re: EA using Wine and ODBC
« on: July 02, 2020, 05:35:04 pm »
If you are willing to invest in EA, then just buy crossover from codeweaver. It's a managed wine system, from guys who are working with Steam for Linux/windows cross platform functionality. It has 'out of the box' recipies for installing EA. It even has paid for support, and is well worth the investment for wine/windows regardless of EA. Try the free demo first.

With it, I have been able to use EA on Kubuntu, centos, fedora etc.

No one with any ideas??
Yes. I have been thinking about this. The default HTML template needs to call the 1990's and ask for its web design back.

The basic problem for an architecture repository is that it is like LDAP. For every change/write you have 1000 reads. So you have to optimize for reading.

There is a "pure HTML5" client side wiki tool called tiddlywiki. It is a *single* html file. If you want to upadate it you neeed to resave the html file.

With the HTML export you can create a tiddlywiki file that looks something like this:

[tiddlywiki code]
<-- EA generated tiddly pages -->
[tiddly encoded pages generated by EA]
<-- end of tiddly pages -->
[end of tiddlywiki code]

This results in a single HTML artifact that has searching. embedded&linked documents, etc. Play with tiddlywiki and you will see what i mean. It is a powerfull way of making a portable linked & searchable EA model.

If you go this route, get in touch as i would like to help. If not, and no one else tries/triggers, i will probably publish something about this in 6 months or so.

i have been experimenting with is using a private git repository for transfer. Git compresses the XML, only transfers changes, and ensures transfer integrity. With gitlab or similar, it also minimises the bandwidth use over a company VPN.

I have been playing with this approach for a working from home model diff and merge process, its only working for a single user though.

There is some kind of horrible manual hacky way to incorporate customized XSLT import/export scripts in an MDG Technology.

Never tried it, but it sounds like it's the kind of thing you're after maybe?

Seems worth checking out. XSLT has its own headaches though. Is the built in XSLT editor worth looking into?

When i poked around in the EA database definition, It's a VARCHAR(50), i was under the assumption that the UUID just needs to be unique. You could probably use that out of the box, if your key is unique, and not have to jump through hoops....

Do you know about Alias? I cannot find any reasonable documentation on it, but i always thought of it as the out of the box support for external Unique IDs. Using alias you can have an element name and unique name, and it is supported in the element view.

In the case of Keywords I'm not aware of any built-in features which use it as of now. But that might change.

I have been wondering what it was for, as we have tags. I always assumed it was for use with aliases to make them unique.

If I needed to do something like that I would probably write a script to export the model according to your custom format.

Others who are more familiar with XSLT might choose for an XSLT transformation from XMI (or one of the other standard xml formats) to your custom format.

Do you mean the built in XSLT tool? Can it work as part of an XML import pipeline?


If it's open source then you could indeed simply modify the code to convert the diagrams starting from package

Do you know where the source for it is?

It was investigated briefly, but before we got any kind of result we abandoned it as inappropriate for what was being done.

Why is it inappropriate? Having a cross platform mechanism to embed external complex code inside an MDG file i would have thought would be in line with EA and Prolaborate.

My tests of the JS engine with Web assembly ay its availalable, and runs for local javascript - because it is an inherent part of the modern JS engine - used as part of its JIT technology too. The problem i had was that no easy mechanism to load an externally compiled webassembly file. I was playing with the idea of using something like a pandoc webassembly port to convert markdown (and other formats) to EA RTF and back.

Before, it hadn't  been updated since around version 3.

What isn't working?

The ability to load webassembly files from a filesystem location.

General Board / Re: Making EA friendly for new users
« on: March 26, 2020, 05:14:51 am »
Does anyone have a guide to ‘Making EA friendly for new users’ ?
There are lots of bits of EA which can be turned off, and some things we can setup for them, which can make the experience for new users simpler and less frightening, but I have never seen them all in one place.

Maybe this thread would be a good place to put such a list? (If people are prepared to add ideas, then I will promise to present it at the next EAUG in the UK)
My starter set is:
1.   Reduce the modelling languages available: Specialize/ Manage Tech / un-tick everything, then just tick any that users need. Maybe even remove them from the list… I see loads of users where everything is ticked.

2.   Remove any top-level menus/ribbons that user don’t need, such as the ‘Simulate’, Construct and Execute menus: Start / Workspaces / My ribbon sets / un-tick the ones you don’t want, or even sub-menus you don’t need. Many users never use these areas,  and it's just more visual clutter, and more places to look.

3.   Give users an initial diagram, which is their default, when will appear when they first login, showing them where to find more information. Start / Workspaces / Model places…

These are just my top 3 - any more out there ?

0. Make a workflow that makes sense to you. Then document it, as *you* will be supporting it!
1. Turn on model security to enforce users and role permissions
2. Define portals and workspaces per role.
3. Make workflow documentation about your role specific implementation.
4. Get Geerts EA validator and tune it to deliver consistency reports on your workflow.
5. Make some reports on your model with the status/states that you would like to see, on the assumption that a bunch of untrained users will stomp all over your interpretation/model/ or EA verterans will make their own (for them superior) submodels (because of course they know better than you, and they can)
6. Roll out access to your model with training about the above
7. GOTO 2. or 0.

This is my current understanding of the workflow.

Any/all of the workflow defaults in EA *must* be considered as undocumented/undocumentable/unsupported pseudo-examples designed to lure you in with the assumption the documentation for them exists somewhere. It does not. Its not for nothing that every guide i have ever seen about delivering/supporting EA for users has a first step which is something like 0-3 in the workflow above.

If there is a Sparx document on these features (that makes sense, and is not an EA based publish->dump of the manual) i have yet to find it.

While i have no interest in, links to, idea of, care for, stock in, or bad words to say about "adepreter's libnaf" (or any of the other EA based off the shelf models like arc42) - they at least are well documented in their use and have user support - because they have already done steps 0-7.

General Board / Re: Export diagrams to PlantUML
« on: March 26, 2020, 04:45:50 am »
did anybody find a solution?
I seem to remember the Arc42 pproject on github had some scripitng to do with EA and plantuml.

Might find something there.

Ditto - I wouldn't recommend that approach.
If you are working in the Enterprise Architecture level then use ArchiMate for enterprise strategies, roadmaps, future states etc.

If you are working on system or software requirements use functional and non-functional requirements the built into Sparx EA. Alternatively use user stories or use cases with scenarios, constraints etc.

Plenty of books out there on the topic.

I've been playing with the idea of a requirements model that starts with Archimate and drills down to UML, while using the native 'otRequirement' object type in the element for the properties window dispatcher. This is based on the assumption that stereotypes can be applied to other EA object types when making an MDG.

Pages: 1 [2] 3 4 ... 11