This Question is Answered

1 "correct" answer available (5 pts) 15 "helpful" answers available (3 pts)
3 Replies Last post: Feb 8, 2008 10:25 AM by phil

Curl source file metadata for SCURL files

Feb 4, 2008 12:38 PM

Click to view rshiplett's profile MVP rshiplett 29 posts since
Oct 17, 2007

As we head down the road that leads to Eclipse+SVN or Eclipse+CVS, it may be a good time to equip our SCURL files with explicit source meta-data

I am thinking of such things as

{source-metadata

copied-from "my-prior-filename.scurl",

previous-char-encoding "windows-latin-1",

previous-package "dragged-in-from-pkg",

last-successful-pcurl "Feb 29, 2008"

}

as our IDE can equip us so readily with this metadata and if it is not there prior to going to plugin status, it is not likely to be in that eventual plugin.

The current mismatch between file-content facilities on the eclipse TEAM menu for CVS and SVN should be a stark reminder of life-under-eclipse ...

Click to view phil's profile Curl phil 28 posts since
Oct 17, 2007
1. Re: Curl source file metadata for SCURL files Feb 4, 2008 2:33 PM

This is an interseting idea, and it's true that the IDE has knowledge about some of this state. The only part that you mention that we already support, sort of, is tracking char-encodings. Since char-encodings are stored as meta-data directly in the scurl file (in a curl-file-attributes clause), you can use your regular source code control history function to track changes to encodings.
Click to view rshiplett's profile MVP rshiplett 29 posts since
Oct 17, 2007
2. Re: Curl source file metadata for SCURL files Feb 5, 2008 8:38 AM
in response to: phil
I am doing some refactoring across multiple projects at the moment and am more often in the IDE than I am in Eclipse/SVN

What if this were just an expression which corresponds to a method on a URL when that url is an SCURL source file

We have something like this when the url is an scurl file with a herald for {package}

Source code versioning systems are so very different in their approaches that I am hoping for a Curl-esque option before we arrive at Eclipse ...

A CURL file has an {applet} herald and an XCURL file has a {script} herald but a source code file is very dependent on a mention in an {include} in a {manifest}

Comment out one line in that {include} and suddenly that SCURL starts to look very sans-metadata

One option that I would like to see in {source-metadata} would be

copyright= "Our_Company"
At the moment our IDE lacks the kind of ClassBrowser that you find in say, the IDE for Extended Object Tcl (XOTCL IDE) or the Smalltalk-style refactoring browsers.

Because our SCURL files can have publicly assigned expressions, public procedures, proc-types, text-procs and classes and macros and now {def}'s it would be great to have

classes-declared = {class-list My-Class_01, My-Class_02}
because of how this can be leveraged by a tool when looking to refactor. But perhaps

classes-instantiated =
might be more than I should hope for ;-)

But otherwise I have to resort to something of a kludge to pull this information from source by brute force, which is not very Curl-esque, at least to my taste.

The temptation will be to make such things a "feature" of an Eclipse plug-in instead of metadata that any programming environment could leverage.

Of course, seen from the viewpoint of packaging, everything you need is already there. Until you are pulling those packages apart in a major refactoring effort ...

And seen from the perspective of {pcurl-file} everything is there - but getting it there once is easy. Getting it all re-packaged another day can be challenging.

Thankfully we have a great IDE already, a terrific Profiler, users who ask for the moon ... ;-)

We could, of course, just have a report available as a pop-up from the IDE 'definitions' pane ... or just grep, gripe, egrep

Or an alternate file view which is an SCURL file view and not a Package view. Before I ask for an XCURL view ;-)
Click to view phil's profile Curl phil 28 posts since
Oct 17, 2007
3. Re: Curl source file metadata for SCURL files Feb 8, 2008 10:25 AM
in response to: rshiplett

There's not much I can say at this point about features in future release, but we are working on adding refactoring support in the IDE. Such a tool will require much more detailed analysis of curl files, and the data will have some information that you're interested in.

It's true that the current IDE has enough information to create a smalltalk-style browser, although we don't have one right now. Maybe in a future release..