On 11/01/2015 01:49 PM, Scott Kostyshak wrote:
On Sun, Nov 01, 2015 at 01:36:53PM -0500, Scott Kostyshak wrote:
On Sun, Nov 01, 2015 at 11:52:08AM -0500, Richard Heck wrote:
On 11/01/2015 11:37 AM, Scott Kostyshak wrote:
On Sun, Nov 01, 2015 at 11:32:17AM -0500, Richard Heck wrote:
On 10/31/2015 05:47 PM, Scott Kostyshak wrote:
The list of tests at the bottom of this message pass on current master
[1] but do not pass on current master after exporting the .lyx files to
2.1.x format. So this is in some sense a compilation test of roundtrip
lyx2lyx.
Is what this shows just that exporting to 2.1.x format and then re-importing
to current format doesn't produce the same file? If so, then that isn't
itself a sign of any problem, since lyx2lyx almost never produces the same
file on roundtrip. This is because reversion often uses ERT.
By "compilation test" I mean that it is a test of whether the LaTeX
compiles with error. So each item on the list compiles without error but
after exporting and re-importing it compiles with error. This is not a
good test because it will miss many bugs, but it seems to me that a
failure does suggest something to look at.
OK, well, that is more of a problem. Do you have any automated way to figure
out which step is causing the problem? One way to do this would be to export
to 498 and re-import, etc, and then see where it fails. Then we would at
least know which reversion-conersion sequence was the cause of the problem.
OK I will look into it and let you know.
497 is fine, 498 produces an error. One thing that is strange is that
after exporting to 498 in the .lyx file there is:
\lyxformat 474
I am using the command
/home/scott/lyxbuilds/master/repo/lib/lyx2lyx/lyx2lyx -t 498 EmbeddedObjects.lyx
> temp.lyx
The problem is that EmbeddedObjects.lyx was already format 498, and
lyx2lyx seems to do the wrong thing when asked to convert it to the
format it already is, i.e., to convert to 498.
I've committed a patch that simply does the null conversion in this
case. As I say in the commit message, this seems better than aborting
with an error.
Richard
