Sparx Systems Forum
Enterprise Architect => Suggestions and Requests => Topic started by: Graham_Moir on March 25, 2010, 09:51:39 pm
-
The Sparx Systems Japan product, RAQuest, has a nice feature whereby in MS-Word you can highlight some text, right-click on the mouse and choose from the context menu to register the highlighted text as a requirement in RAQuest (and therefore in the EA model you are working on as RAQuest works on top of a standard EA model). If you highlight 2 paragraphs in Word, the first is used as the name/title for the requirement element and the second populates the notes.
Does anyone agree this should be a core feature in EA rather than in an add-in ?
-
Nope. For managing requirements I would stick to a special requirements management tool. I would expect from these tools to export a valid XMI so I could mirror the requirements to EA where I could go forward with traceability to use cases. I haven't really used RaQuest (except for a short time when it was first in beta) but I think it's not a fully fledged RM tool. It's something for the low-medium range. It helps handling requirements but it's missing a lot like e.g. impact analysis.
Further: why should EA be integrated directly with that MS-Scrap? It would be more vulnerable and dependent and cost-intensive. A tool is a tool and an interface is an interface. If you'd like that integration plug in the interface to MS-Scrap. I wouldn't like to see that on my machine.
b.
-
OK, I understand where you stand. :)
Nevertheless, in our case, we have users delivering business requirements written up in Word documents. We want to capture those in EA as granular individual elements so that they can be properly traced and we don't use a separate requirements management tool, so we're left cutting and pasting the information, or writing it up a second time. I just felt that an equivalent function to the one RAQuest provides would cut out a lot of the overhead of this process. If it was there, it wouldn't hurt and you could still follow your XMI process
-
How about the following: Create a copy-to-csv utility from within Word which just appends selected text to a csv. Then import this csv into EA creating a bunch of requirement elements.
b.
-
Well yes that would work, however it is a lot more intensive than, 1) highlight the text in Word, then 2) register that selection as a new requirement in EA - job finished.
What's more the functionality exists today - it just isn't directly in EA.
-
Try this in version 8.
1) Create a Requirements diagram in EA.
2) Highlight the the text in Word
3) Drag and Drop the text onto the diagram.
4) Select Requirement.
-
We are using RaQuest as a front end to EA for requirements management. While you can certainly do RM work in EA, the more graphical approach is less effective for many people (until you start working on traceability and further analysis work in the models).
Having worked with Rational ReqPro and Borland Caliber, I would say that RaQuest is not bad, especially for the money (about 10% of the cost for 80% of the functionality). It does most of what I want (including a form of impact analysis now), places the data natively into EA (where I want it anyway for traceability) and provides the users with a "table based" UI that they are more likely to find approachable than the full complexity of EA.
The good thing about Sparx is that they focus on the core functionality and provide an extensible platform for third party developers to provide functional add-ons. I think this is better than trying to be all things to all people, and ending up with bloatware containing functionality many people will never use.
It's always nice to see Sparx add new stuff (I must try out Simon's example), and the ability to extact text from Word into any element is always going to be useful. But I am very happy with RaQuest, and recommend it.
Just my 2 cents, but hopefully it will be useful.
-
What Simon has described looks as though it will meet my requirement - I have to give it a try.
Regarding RAQuest, we're not against it and we have some licences, it's just that it's quite expensive in comparison with EA itself and we have preferred to invest our budget in additional EA licences.
G.
-
It looks like the function Neil mentioned is still quite beta. I dragged text from Notepad++ and selected Object as target. On the diagram the text was visible in the name but a dbl-click didn't show it. Though I think this feature is very useful and should fulfill Graham's needs (as soon as it works as designed).
One remark: it would be nice (as Graham stated) if there is a line break in the text that the first line goes to the name and the rest to the notes.
b.
-
Just tried Simon's suggestion.
I opened an MS Word Document and highlighted a body of text.
I dragged the text onto a requirements diagram and selected "Requirement". It created a requirement element but pasted the text somewhere in the Tagged Values tab instead of Notes.
Looking forward to the complete solution.
Segun
-
I welcome this new feature as a step forward, but a "complete solution" will require a real interface to an external source document so that changes in the source document will result in changes in EA. (This is the idea behind the DOORS add-in.)
-
I agree. I also anticipate a solution that goes further by parsing the text and using some clustering algorithm like K-means to cluster the Nouns and deliver a list of candidate business concepts for a domain model.
-
A further thought on this...
We should be able to trace changes to individual requirements, which means we need to be able to track individual requirements in the source--even if the requirement moves in the source. What this means is that each requirement needs a unique identifier in the source. Hence I think simply typing requirements in a word processor, without identifiers (and a tool that can read these), is not sufficient for requirements management--if theword processor document is to serve as the requirements source. (Mind you, that is usually what happens in our office, much to my dismay.)
-
Paul, you're obviously right since tracing to the source is essential. OTOH this is not a task which can be covered by EA - this is simply too far off. A good RM tool must be able to handle that. It's a very specialized task. And it can not be that EA also covers this. EA is already a swiss army knife for the UML modeller (like I use to say in my lessons) but if you're going to also use it as a hammer and as workbench it will dfinitely be overloaded.
b.
-
Well, I was more concerned with someone relying too much on text documents. Traceability throughout a project is actually quite challenging, I've found.
One approach is to write requirements directly in EA. In particular, SysML is designed to support this and I think the concept is outstanding. (One ends up with only one model with precisely defined relationships between the elements rather than a set of documents containing elements with difficult to navigate, if defined, relationships.) In the long run this is where I would like to go. On the other hand, the traceability capabilities in the current version of EA are not, I think, sufficient, and not on par (yet) with those of a good requirements database management tool. Specifically, we need a way to show effiiciently that we satisfy requirements with this design, that code, and these tests. (The Enterprise Tester add-in is a big help towards this last effort.) In EA (currently) we lack the capability to show requirements coverage effectively. In particular, EA does not even distinguish between header requirements and detailed requirements when dealing with requirements satisfaction. Since SysML is very much in the EA purview, I think it is not at all illogical for to pursue traceability capabilities further in EA.
-
For the basic functionality that the discussion in this thread started on, I have submitted a v8 beta bug requesting refinement of the functionality described by Simon so that title/notes of the target requirement element are populated.
-
Now that version 8 has been released, I've given this a try.
It looks as though nothing has changed though. Using Simon's suggestion
- the requirement gets created and the highlighted text is displayed as the title of the element (but that's not too helpful with a paragraph of text)
- the text gets buried in the tagged values, I can just about see it, but I can't find a way to edit it
- when the properties dialogue is brought up the title/short description is the default e.g. "Requirement3) and the notes field is blank :(
For me, while I understand the argument for full traceability and integration with source documents, this current functionality is so close to being what I would like right now, but at the same time so far away, that it is very frustrating !
-
I've raised this formally as a v8 bug.
-
While the functionality is neat it has two serious flows:
1. After dropping text it seems to be stored in some strange place. It is not a tagged value it is a name but it cannot be accessed in any reasonable way.... until you try to edit requirement in project browser (pressing F2) and then you hit OK. From now on requirement name is reflected in properties.
2. For such functionality it doesn't make much sense to copy the text into a name field that has 255 chars limit, you could take either first line or 80 chars as name and put rest in notes. Otherwise the part of the text is gone. (I should also point out that context menu of such long item spans over 1500 pixels in width :) )
BTW does it help a bug to be reported many times? Should I report it too or will your report do?
Regards
Kris
-
...
BTW does it help a bug to be reported many times? Should I report it too or will your report do?
It seems so...
Sparx (from previous postings on related subject) appear to take no cognizance of support in the user forum.
Only formal support via the formal mechanisms (links below) are taken into account.
It's entirely too tedious (more EAUI), but that's the current functionality.
So report away... - just add a note here that you've done so otherwise there'd be NO visibility that there has been support for the bug report.
Paolo
-
Kris is correct regarding the 255 character limit etc.. and I've sent in additional information along these lines in the bug report.
The suggestion - included in the report - is to make this work like RAQuest, in other words, if there is only one carriage return/line feed, the text is added to the Notes field of the element, otherwise if there are multiple cr/lf then use the text up to the first as cr/lf as the title/short description, and put the rest in the Notes.
-
FYI - Sparx have come back to me on this. The issue as logged has been split into 2.
1) Name/Short description not populated correctly - Bug. Reference 10045519
2) Suggestion to populate the Notes field or Title/Description AND the Notes field depending on the number of carriage returns. This is being treated as a feature request, which I suppose it is, although I would argue the current approach is a design bug in fact - anyway no reference provided.
If anyone feels strongly about this functionality, then I suggest you also log the issues with Sparx with the usual intention of trying to create a weight of opinion about it.
-
Did that work in any case?
I'd be happy to see this implemented but have absolutely no insight in the design process driven for EA. I doubt it's a lotto machine. Or just some of these marketing thingies.
b.
-
There are a couple of lines in the version 9 release notes that suggest this might have been addressed !
So perhaps the guys at Sparx do listen sometimes :)
======
Creation of Elements by dropping text onto a Diagram improved
- Now supports text and note elements, by setting the Notes field instead of the name
- Multi-line text will set the name to the first line, and notes to anything following that
- Opening properties dialog before reload will now show the correct name
======
-
I've tried this in the released version of EA 9 and it works pretty well - along the lines that were suggested in the bug report/feature request. Only one additional thing I would add which is to replace non-printable characters such as tabs with spaces, otherwise job done.