Re: Dia find & replace?

[Johann Tienhaara <jtienhaara yahoo com>]

Thanks Hans!

If anyone besides Hans is reading, I have a question for any interested Dia-zens about the find & replace "checkboxes" below...


--- On Fri, 10/17/08, Hans Breuer <hans breuer org> wrote:

> From: Hans Breuer <hans breuer org>
> >> I made a few changes to find-and-replace.c so that --
> >> optionally -- any and all text properties can be found &
> >> replaced (rather than just the "name" and
> >> "text" properties of an object).
> >> This gives me the rest of what I was looking for -- the
> >> ability to replace text in UML attribute types and operation
> >> parameter names and what-not.
> >>
> >> I would definitely appreciate your feedback.  
>
> First thing is keeping consisteny , at least on the single
> file level. You
> have introduce some new programming style, which make the
> patch unnecessary
> hard to read:

My apologies for the inadvertent stylistic shifts.  I was actually making a conscious effort to follow the Dia style.  But I didn't do one crucial thing: proofread the patch line-by-line for style-sloppiness!  I will certainly do that in future.


> >> Especially
> >> since some of the data structures had me scratching my head
> >> a bit (like the list-of-property-lists "records"
> >> inside sarray and darray properties -- why a list-of-lists
> >> of Properties? Why not just a list of Properties?).
> 
> This on is simple:
> For example the UML class has methods (list).
> A method is a _record_ with name, return value and
> parameters (list).
> A parameter again consists of a record with a name, a
> (default-)value, a
> type etc.

Thanks for the explanation of sarray & darray.  If it's not verboten I might add a comment to the header the next time I get a chance for some hacking, just so that your explanation is in the same spot as the structure def'n.


> >> I have not yet touched the i18n/l10n stuff either.  There
> >> is currently a rather cumbersome message for the "all
> >> properties" checkbox.  I'd rather have a
> >> terse-yet-still-meaningful message before I figure out how
> >> to hack the message catalogs.
>
> No need to hack any message catalogs. By simply marking a
> translateable 
> string with _() your are usually done with
> internationalization. 
> Localization is than done on the message catalog by the
> translation teams.
> ...
> The thing missing is a good name for the two new options.

If anyone who's interested is reading this, there are a few checkboxes in the "find & replace" dialog:

  [ ] _Match case
  [ ] Match _entire word only
  [ ] Match _all properties (not just object name)   <-- ***

The last one is obviously way too clunky...  Any suggestions?

The basic idea is that if you have, say, a UML Class object:

  Class Foo
  - Attribute parent: Foo
  + Operation get_parent(): Foo

Then when "all properties" is NOT checked (default), you might replace "Foo" with "Bar" and end up with:

  Class Bar
  - Attribute parent: Foo
  + Operation get_parent(): Foo

Whereas when the third checkbox IS checked, you would end up with:

  Class Bar
  - Attribute parent: Bar
  + Operation get_parent(): Bar

I have no idea what to name the checkbox...  Any suggestions?

(Or completely different ways of looking at the problem would be welcome too!)


> I've doen the reformatting and some minor changes and
> just commited it.

I noticed you added a "FIXME" which I didn't consider:

  //FIXME: do we realy want a replace all here?

It may require some uprooting of code to fix that (stopping after the 1st property matched -- though I believe the fix belongs elsewhere in the code than where the comment is...).   I'll certainly take a crack at it -- probably won't be for at least a few days though.

I also seem to have left in one (of 2) "TODO"s which I should have dealt with -- string list properties.


> [1]
> http://www.thomascrampton.com/internet/netidentity-email-outage-19-hours-and-counting/

Oh my...  Have you ever considered directing your inbound email to "hans example com"?