Alex Elliott

The internet home of a prospective software engineer

This is my personal blog where I discuss projects that I'm currently working on, work I've recently completed, or write about any topic which has caught my interest in the world of Computing from my studies or from my personal research.

Latest Articles

Back to Work

July 29th, 2010

Well, been a while since the last blog post I’ve written, and a similar stretch of time since I really got much work done on ExpressionEditor. But as you might have guessed since the blog is winding back into life, so to is my work on EE.

Changes So Far

I’m still working my way slowly through the todo list for EE (updated now on the wiki.github home page for EE), major differences that I’ve already managed to cross off include adding support for the ICU regular expression library, and migrating the build process over to CMake (which is still partly ongoing, the more people build it on as many platforms as possible the better – please do report any issues, in IRC would be favourite as I can work through the issues directly).

ICU (International Components for Unicode) supports a regular expression engine that seems to be a popular choice particularly for Mac-based programming, and appears to be fairly full-featured.  I hope that ICU support will be useful to people, and that it gets good use.

As to the migration to CMake, this is for several reasons: it should make it easier to distribute at the moment it supports a very basic “make install” target, and I will be expanding that to bring the common files into /etc (and at some point I should improve the content of those common files, but that’s something I can do later) and also expanding that to take the README, CREDITS, and similar files into documentation.

It should also now correctly auto-detect which of the optional dependencies (all of the regular expression libraries bar that included in Qt4.6+) are present on the system, and build a version of EE which supports the ones which are available.  This hopefully will make it less of a pain for those who don’t have PCRE or ICU (or in the case of Win32 POSIX’s regex.h).

And talking of making it easier for Win32, after a fairly long period where no EE was compiled on Win32 it was recently built, at the least proving that the code is still very happy to compile there when it’s not missing optional deps – which is very good news to get. :-)

Looking Forward

Of course, there’s still quite a lot to do to reach the plans I’ve set out, and since I’m working full-time I don’t have as much time as I’d like to work on EE (which I hope you’ll agree is an interesting and quite fun little tool, and it’s just as nice to work on :-) ).  But I expect to make some progress nonetheless, and if I do, I’ll keep all one maybe two people who have this in their RSS readers up to speed. ;-)

Expression Editor Update (2)

January 24th, 2010

Since I’ve had some more time to work on Expression Editor recently I thought it was about time I wrote another update for the progress of the project, and some related news that affects it.

Expression Editor on Mac OSX

A recent screenshot of Expression Editor on Mac OSX

From Last Time…

In the previous post I noted a few areas in progress and some that I wanted to look at in the future.  So to catch up there, Drag&Drop is generally a bit more reliable and produces slightly neater results but is otherwise unchanged so far, and the new testing widget is still waiting.  A significant change has been made in the area of supported regular expression formats however.

The application now has backends for Qt4, PCRE and POSIX ERE formats (though the visualisation could still mess up some PCRE/POSIX elements, let me know if anything breaks).  You can select the format you wish to work in from the menu bar, it will be displayed in the bottom right of the screen so you know which mode it is currently in, and the save format has been slightly extended to save your preference for each particular expression.

The default mode has also been changed to PCRE, since it is probably the most powerful backend available.  Another minor UI change has been included which is an expression status indicator to the right of the text input.  A green tick while valid, and a red exclamation mark when invalid, in addition if you mouseover the invalid indicator, the tooltip is the error returned from the active regular expression backend.

In Related News

As you probably saw above the screenshot used is from Mac OSX.  In order to improve my capacity to test Expression Editor I’ve gotten myself a Mac Mini as well as my Slackware Linux laptop.  Set up with Synergy+ this means I can simultaneously develop the application in Linux and test it in OSX.  One behavioural difference between the two operating systems has already been resolved, so hopefully the application should start behaving much more reliably on OSX as well as Linux from now on.

Expression Editor Update

December 24th, 2009

A fair bit of progress has been made since my last blog entry so I thought I’d note a few things that have landed in the repository and a few things that I intend to add at a later date.

Drag and Drop

Initial support for drag and drop editing has been added.  You can now re-order the elements of the expression by dragging an element in the visualisation to one of the valid drop zones (which are automatically highlighted as you can see in this screenshot).  With this in place it becomes significantly easier to add the other bits of drag/drop editing I want the editor to support.  Eventually as well as reordering (plus the double-click edit dialogs which are also currently included for several elements) I aim to include:

  • Drag/drop adding of new elements from the toolbar to the left of the visualisation.  This should probably spawn a dialog/wizard and then insert the resulting regular expression element into the current expression.
  • Reordering needs more support in the alternatives item, currently there are only valid drop zones to place items inside current alternation branches, and there should be a drop zone allowing the user to drop an element in as a new alternative.
  • Possibly a simple “trash” element, which simply accepts the drop, and results in the item being deleted from the scene.

Regexp Formats

As stated in a few places in the application, before the initial release I hope to support PCRE, POSIX Extended and Qt format regular expressions.  This means supporting a range of different regexp syntaxes, and intelligently warning when switching between formats if some of the expression cannot be used directly in the new format, it should also offer to try to translate the expression if such a problem exists.

For example, if we’re currently in PCRE mode and we have an expression containing “\w” and we switch to POSIX Extended, this should trigger a warning and then offer to translate, turning “\w” into “[[:word:]]”.

At the moment, the application only supports Qt’s internal format, and I think correctly represents much of what it supports internally.  The format is very much  like a slightly restricted PCRE format, so Qt/PCRE conversion should be fairly straightforward.

Expression Testing

The editor currently has an element at the bottom of the layout which allows you to test the regular expression for given short strings.  This is good for most cases, since it allows you to have a few regexp “unit tests” of sorts, where you test fringe cases and observe if it matches, partially matches, and whether the capture groups work as expected.

In addition to this it would be useful to have a few other methods of testing included.  The testing widget should eventually be a tabbed widget with the currently available tester as an option, then also having at least two additional panes.  A “bulk text” pane which  takes paragraph or longer inputs of text and highlights all instances of that section which are matched by the regular expression, and a “replacement” pane which allows you to input a similar length input to “bulk text”, and apply the regular expression with a given replacement string (which could also be a regular expression).

Anyway, that’s what I’ve been working on and some of what I want to include later.  Work goes on. :)