1306 How may I avoid nonsensical fragmented alignments...

[chrisjj]

How may I avoid nonsensical fragmented alignments such as this:

http://img808.imageshack.us/img808/2...etextcompa.png

This is with Line alignment Standard method Slow and thorough. But Alternate method is no better.

Thanks.

[Aaron]

Hello,

Thanks for the screenshot. We'll need a pair of example files to look into this alignment and add it to our test cases.

Please email us:
1) The full line of text from both sides as a pair of example files. We don't need the full file if just that line reproduces this alignment.
2) Your current Support.zip from the Help menu -> Support; Export
3) A link back to this forum post for reference.

[chrisjj]

Done. Is this WAD?

[Michael Bulgrien]

I've reported some annoying alignment issues with the current algorithm as well... As far as WAD goes, I think it was a trade-off. The algorithm was chosen for its speed/performance at the expense of a few quirky alignment issues and seeming inconsistencies when differences exist.

[chrisjj]

The algorithm was chosen for its speed/performance at the expense of a few quirky alignment issues

One would hope to get the "Slow and through" offered.

[Zoë]

One would hope to get the "Slow and through" offered.

The alignment options only affect line-by-line alignment. The character alignment is not configurable.

[chrisjj]

Oh. Thanks.

And I guess no tradeoff control is needed for character alignment because that is such a less intensive task. Making me wonder why it is so much less thorough that other programs'.

[Zoë]

It's not less thorough, it's different. The algorithm we use produces an "optimal" alignment, where optimal means the largest number of matching characters. The problem you're seeing is that it doesn't give weight to grouping adjacent characters, and that's something I hope to fix in a future release.

[chrisjj]

It's not less thorough, it's different.

It less thorough than e.g. the MS Word of ten years ago that in this example has no problem aligning the word activies with itself.

it doesn't give weight to grouping adjacent characters,

Note: it goes haywire only if there is a preceding mismatch:

and that's something I hope to fix in a future release.

Good to hear. Thanks.