On 2015-12-04, Kornel Benko wrote: > Am 4. Dezember 2015 um 01:40:25, schrieb Uwe Stöhr <uwesto...@web.de> >> Am 03.12.2015 um 18:45 schrieb Kornel Benko:
>> >> Strange, no relevant change recently. Does zh_CN/Intro compile with TeXF? >> > >> > They compile now. After reverting some 'unneded' removed code. >> Why do you revert this? I can compile the file here without problems and >> that a \clearpage is necessary for you to compile it seems strange to >> me. Maybe you found a bug. Now you reverted and we don't investigate >> further. > Dear Uwe, you are not alone. I could not compile. So for me the > previous state was better. So (with the "unneeded" code cleaned up) we have an instance of "unreliable": works depending on some external factors (package versions, configuration, phase of moon). These are always the most problematic cases, because the true nature can only be detected via communication between the parties... Proposal for a general guideline: * If a change breaks compilation for the target format (for the manuals pdf2) without solving some important other issue -> revert the commit (This is what Kornel did.) * If a change breaks compilation for some non-target format (for the manuals everything except pdf2) -> invert the test. > But you have a point, we should investigate. My proposal here would be to keep the "unneeded" code unless we established that it is really not needed on all places but add a LyX-Note explaining this in the document: The following clearpage is required to avoid compilation errors at some installation for an unknown reason. This should help to keep the issue in memory¹ while preventing frustration for users failing to compile the manual. ¹ a bug-report helps even more. Günter
