Q&A tool glitches

[Marcus Wagner]

Original ENVY Q/A, as important as it is, suffers from ancient glitches, not fixed since ages.

1. CcHighlightingWithPositionComplaint treatment is incomplete in the high lighters method findHighlightPositionsUsingStringSearch. 
So in cases with several complaints in a method, high lights are mishandled and their navigation is missing (next, previous).

2. Possible automatic repair is not offered for several cases, e.g. unnecessary temporaries, missing yourself, compiler warnings, public/private inconsistency and unnecessary brackets.

3. Q&A of several abt generated methods potentially raises complaints which usually should not be treated at all (so to not disturb abt mechanisms in the background). A convenience method to select abt generated methods and then to remove their complaints would essentially speed up their treatment, but was not offered.

4. There is a (rare) architectural prerequiste problem if three different applications compete in changing / applying extensions to a root application. 

In Q&A such a prerequisite problem result in visual conflict between the root ETTools, CCUserInterface and the important enhancement of Scintilla support which could have been resolved, but was not: 
EtTTools offers browser framework with a code pane,

Q&A uses the code pain to display non code in it (description, statistics) and 
Scintilla in turn applies colored formatting the content of the pane. 

Not to say that mixing up was never resolved: code critic results should not be Scintilla formatted if content is no code at all. Such cannot be resolved in prerequisites, as any load combination is possible and allowed for in pricnicle independent applications.

5. Some complex code patterns are detected, but their repair is tedious. Automatic treatment would be helpful. This leads to a series of complex code refactoring items, beginning with missing cascades...

6. For financial applications: accidential use of float instances (instead of ScaledDecimals) was an extreme risk. Q&A checking for float instances helps here, but was not offered.

7. Broken code associations (pool definitions) was a classic domain of Q&A, especially in context of packaging. This became out of scope, in particular due to the high quality handling provided by the tools. It became nearly impossible to write and save something which ends up in a defect compiled method.

All these topics led to several Q&A enhancements and fixes from which I personally profit. 

I do no know if this is of interest for anybody else and how to provide this to the public.

-

M

---

I 

[Esteban A. Maringolo]

Hi Marcus,

Thank you for taking the time to report these issues and suggested improvements, as well as the STS one reported in another email to the list.

We're taking notes and we will create development cases for them, so we can implement them for inclusion in future releases of VAST.

Best regards,

Esteban Maringolo VAST Consultant Senior Software Developer  emari...@instantiations.com  @emaringolo  /emaringolo  instantiations.com

--
You received this message because you are subscribed to the Google Groups "VAST Community Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to va-smalltalk...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/va-smalltalk/70545501-87c4-4901-a21f-f50d07fcb4adn%40googlegroups.com.

[Marcus Wagner]

I found this tiny fix in my repositories: disabling Scintilla when Q&A is showing something else than code, causing illecit formatting now. This turns it off:

CTResultsBrowser>>stsIsShowingParseTreeFriendlyText
"Return true if the classification of text being shown
 is suitable for building a parse tree"

"Avoid highlighting of descriptions when Scintilla is desplaying the code pane"

^(self textCanBeAccepted) and: [super stsIsShowingParseTreeFriendlyText]

Hope that helps.

-

M