> (Sorry, can't find June's solution in jwiki. Suggest search Pipermail.)
Sorry again, here it is… http://code.jsoftware.com/wiki/User:June_Kim http://code.jsoftware.com/wiki/File:Box.ijs On Wed, May 10, 2017 at 1:00 PM, Ian Clark <[email protected]> wrote: > Can I offer my two-penn'orth? > > A lot of water has gone under this bridge, but that was years ago, back in > j602 days. It's all gone a bit rusty for me. > > Both June Kim and myself once offered completely different solutions to > the problem(s!) of displaying boxed unicode (<su). June's follows and > extends the actual J algorithm. Mine (~addons/format/sbox) does a > cheap'n'cheerful job of patching (":<su) having computed the boxes with > single-byte placeholders in a surrogate (su). It's value :-)) is that it > emphasises preserving the integrity of the boxes. It might actually handle '�' > – which becomes a valid unicode code-point at some stage. > > Sorry, haven't looked at it for a year or two, don't know whether it still > works. And "works" is a relative term: if you give sbox too many different > code points (>127), it will inevitably fall over. > > Anyway the article… > http://code.jsoftware.com/wiki/Addons/format/sbox > …might get you ahead a month or two when cobbling your own solution, since > it's a taste of the wacky world of utf-8 and what you're up against. (…Or > maybe it will set you back a month or two.) > > The take-home message is that it's highly font-dependent. And font-version > dependent. > > (Sorry, can't find June's solution in jwiki. Suggest search Pipermail.) > > I conjecture a robust solution would abandon box-drawing chars in favor of > a graphics one: rasterizing separate leaves on the tree of (<su), then > using Quartz (on Apple devices) to draw borders round the resulting > collection of rectangles. Rather like a good spreadsheet does. Meanwhile, > unless someone's willing to go that route, is it really going to repay the > effort? Might it not even be better to print 3 stars and a shrug whenever > utf is detected, than output a corrupt display – which sends the message to > novice users there's a (potentially fixable) bug? > > But a graphics approach might even extend to 3-D. And beyond…! Just think > of the visualizations *that* would make possible. > > > On Wed, May 10, 2017 at 3:03 AM, bill lam <[email protected]> wrote: > >> The feature was there because a co-author (bill) was lazy. >> A better user experience would be possible if illegal sequence >> be skipped or replaced by space or other special characters. > > …snipped… > > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
