Re: Aw: Re: #10199: Non-encodable characters with XeTeX
Le 10/07/2016 03:30, Scott Kostyshak a écrit : On Fri, Jul 08, 2016 at 10:06:52PM -0400, Scott Kostyshak wrote: On Fri, Jul 08, 2016 at 09:23:32PM +0100, Guillaume Munch wrote: In that case, if you already have everything set up and that does not cost you any more effort, thank you for stepping forward. I will run them over the night and give the results tomorrow. Tests indicate that commit b170b6e4 lead to ctex_pdf4_texF going from failing to passing, and no other changes in the tests. So according to the tests [1], b170b6e4 is safe. Scott [1] it should be said that about 500 (out of 5000) of the tests are failing, so they do not provide nay information. Thanks Scott for the tests and Günter for the fix.
Re: Aw: Re: #10199: Non-encodable characters with XeTeX
On Fri, Jul 08, 2016 at 10:06:52PM -0400, Scott Kostyshak wrote: > On Fri, Jul 08, 2016 at 09:23:32PM +0100, Guillaume Munch wrote: > > > In that case, if you already have everything set up and that does not cost > > you any more effort, thank you for stepping forward. > > I will run them over the night and give the results tomorrow. Tests indicate that commit b170b6e4 lead to ctex_pdf4_texF going from failing to passing, and no other changes in the tests. So according to the tests [1], b170b6e4 is safe. Scott [1] it should be said that about 500 (out of 5000) of the tests are failing, so they do not provide nay information. signature.asc Description: PGP signature
Re: Aw: Re: #10199: Non-encodable characters with XeTeX
On Fri, Jul 08, 2016 at 09:23:32PM +0100, Guillaume Munch wrote: > In that case, if you already have everything set up and that does not cost > you any more effort, thank you for stepping forward. I will run them over the night and give the results tomorrow. Scott signature.asc Description: PGP signature
Re: Aw: Re: #10199: Non-encodable characters with XeTeX
Le 08/07/2016 10:10, Scott Kostyshak a écrit : On Thu, Jul 07, 2016 at 12:30:45PM +0200, Guillaume Munch wrote: I can do that. Can somebody tell me (again?) which command line will run the appropriate test? The basics are: # In a new directory (preferably outside of the source), run this command: cmake -DLYX_ENABLE_EXPORT_TESTS=ON /path/to/source # then build make # then run tests ctest There are currently many tests failing, so you would need to run the tests without the patch and with the patch, and see if there are any differences. The tests take a long time. You can run them in parallel but this can mess things up. See in Development.lyx the part about "‑‑rerun-failed" of ctests for how to get the advantages of parallel testing without the worry. Because the tests take so long, I would do the build with the patch in one directory, without the patch in a separate directory, and then run two instances of ctest and leave it overnight. I can run the tests if the above seems like too much trouble. In that case, if you already have everything set up and that does not cost you any more effort, thank you for stepping forward.
Re: Aw: Re: #10199: Non-encodable characters with XeTeX
On Thu, Jul 07, 2016 at 12:30:45PM +0200, Guillaume Munch wrote: > I can do that. Can somebody tell me (again?) which command line will run > the appropriate test? The basics are: # In a new directory (preferably outside of the source), run this command: cmake -DLYX_ENABLE_EXPORT_TESTS=ON /path/to/source # then build make # then run tests ctest There are currently many tests failing, so you would need to run the tests without the patch and with the patch, and see if there are any differences. The tests take a long time. You can run them in parallel but this can mess things up. See in Development.lyx the part about "‑‑rerun-failed" of ctests for how to get the advantages of parallel testing without the worry. Because the tests take so long, I would do the build with the patch in one directory, without the patch in a separate directory, and then run two instances of ctest and leave it overnight. I can run the tests if the above seems like too much trouble. Scott signature.asc Description: PGP signature
Re: Aw: Re: #10199: Non-encodable characters with XeTeX
Le 07/07/2016 08:53, G. Milde a écrit : - Ursprüngliche Mitteilung - #10199: Non-encodable characters with XeTeX -- Reporter: andnot | Owner: lasgouttes Type: defect | Status: fixedinmaster Priority: normal | Milestone: 2.2.1 Component: general|Version: 2.2.0 Severity: normal | Resolution: Keywords: regression | + Comment (by gadmm): As far as I understand it makes sense for 2.2.1 and is even very desirable. Only Günter can say whether it is safe. I imagine this requires some tests and maybe adjustments of supplied templates and examples. The patch making this bug "fixedinmaster" is "minimal invasive" and should be save. I recommend backporting to get back the desired behaviour (keep utf8x encoding with Xetex+TeX fonts as power-user option) also for the 2.2 series. (As this is what we had in 2.1.x, too). I can do that. Can somebody tell me (again?) which command line will run the appropriate test? I dont suspect any changes required but running the complete test suite before is recommended. (I am on holiday, offline, for two more weeks) Have fun.
Aw: Re: #10199: Non-encodable characters with XeTeX
- Ursprüngliche Mitteilung - > #10199: Non-encodable characters with XeTeX > -- > Reporter: andnot | Owner: lasgouttes > Type: defect | Status: fixedinmaster > Priority: normal | Milestone: 2.2.1 > Component: general | Version: 2.2.0 > Severity: normal | Resolution: > Keywords: regression | > + > > Comment (by gadmm): > > As far as I understand it makes sense for 2.2.1 and is even very > desirable. Only Günter can say whether it is safe. I imagine this requires > some tests and maybe adjustments of supplied templates and examples. The patch making this bug "fixedinmaster" is "minimal invasive" and should be save. I recommend backporting to get back the desired behaviour (keep utf8x encoding with Xetex+TeX fonts as power-user option) also for the 2.2 series. (As this is what we had in 2.1.x, too). I dont suspect any changes required but running the complete test suite before is recommended. (I am on holiday, offline, for two more weeks) Günter
