On Fri, Oct 27, 2017 at 07:38:32AM +0000, Guenter Milde wrote: > Actually, the patch deals with both, forward conversion and backward > conversion. In tend to summarize both conversions under the tag "backwards > compatibility".
I see, thanks for the correction. > > Indeed, our "Development" manual covers this: > > > While the conversion routine is required to produce a document that > > is equivalent to the old version, the requirements of the reversion > > are not that strict. If possible, try to produce a proper reversion, > > using ERT if needed, but for some features this might be too > > complicated. In this case, the minimum requirement of the reversion > > routine is that it produces a valid document which can be read by an > > older LyX. If absolutely needed, even data loss is allowed for the > > reversion. > > > The current code (without the patch) clearly already satisfies the > > "minimum requirement". > > I don't agree: > > The current conversion routine does *not* produce a document that is > equivalent to the old version: The minimum requirement, as quoted above, is that a valid document is producecd that can be read by an older LyX. I interpret this to mean just that a .lyx file is produced that can be opened in older LyX without parser errors. In any case, equivalence is not the minimum requirement. > * If you used literal em- and en-dashes in pre-2.2 documents, > you must manually unselect > "Document->Settings->Fonts->Output em- and en-dash as ligatures" > to ensure unchanged behaviour. > > > OTOH, the backwards conversion hack for 2.2 adding ZWSP > > * needs to be reverted on forward conversion, > * prevents proper back-conversion to 2.1 and older, > * makes documents uncompilable with LyX 2.0 or earlier > > I replaced the ZWSP with preamble code, but we could also just remove it > (as ERT was voted against and "the requirements of the reversion are not > that strict"). > This would make the patch simpler and less error prone. > > > All this to say: I don't think the patch is too > > important for 2.3.0 and in my opinion I'm fine if we do not put it in. > > Literal dashes may be less common than "ligature dashes" in pre-2.2 > documents, however > * the UserGuide documents Insert>Symbols als method to insert dashes since > several LyX versions, > * there is evidence in mailing list threads that users were inserting > literal dashes via system keyboard shortcuts years ago. > > The problem here is, that we cannot know for sure how large the number of > affected users is: the changes in 2.2 did not affect them and they will only > speak up once the new damage is done. > > > I can propose a medium and a minimal patch: > > The minimal patch just removes the ZWSP hack, > the medium patch also sets the new setting based on content. Thanks for your new proposal. If the new proposal gets support after rc1, I am fine to consider it, and as I said I could spend some time helping with the testing. If other LyX developers are in favor, we can even wait a couple of days to fix this before rc1. I will not push the tag yet (I will tag and tar locally and send the resulting tars to our packagers). So if we decide as a group that there is an important fix we need in rc1, then I can retar and send new tars to the packagers and they can make new binaries. For now, I'm planning to go forward with rc1 unless others speak up, since the only other opinions that I interpret are that Enrico does not think any patch is worth addressing this issue at this point, and that Richard also looked and did not propose delaying rc1. Scott
signature.asc
Description: PGP signature
