Hey, Thank you Tim and Chris for the excellent detective work.. In fact you seem to be a really good team, i observe.
So I agree to keep the hands down until you finished in develop and the n we start and prepare an RC. Julian Von meinem Mobiltelefon gesendet -------- Ursprüngliche Nachricht -------- Betreff: Re: Bugfix-Release 0.3.1 Von: Christofer Dutz An: [email protected] Cc: Oh wow ... Well in that case we should fix this in develop and merge that to 0.3 branch prior to cutting a 0.3.1 (Actually this makes the last fix obsolete or is a superset of it) Chris Am 13.02.19, 17:59 schrieb "Tim Mitsch" <[email protected]>: Hi Chris You have been completly right. I just made a test vs our device where reading an odd and an even number of array-items (based on one-byte base type e.g. USINT). Reading a odd number this resulted in filling byte, whereas an even number did not have a filling byte. I think we should implement this as well (dependent on configuration i could cause some NPEs on reading) and validate behavior by a significant test. Best Tim Am 13.02.19, 16:12 schrieb "Christofer Dutz" <[email protected]>: Hi Tim, I thought of reading an array of 3 bytes ... that should produce an odd number of bytes in the response (Don't forget to request another item after that) Chris Am 13.02.19, 15:55 schrieb "Tim Mitsch" <[email protected]>: Hi Chris I thought about that too, but did not evaluate if something like this can happen. You mean something like odd-adress to even padding?! But the SPS is answering the base types requested isn't it, so there is no basic type that has an odd length in byte except BYTE, USINT and all other one byte long datatypes, or am i wrong. But before preparing BugFix-RC we'll should check this, you're right. Later this day i can support with this. Best Tim Am 13.02.19, 15:42 schrieb "Christofer Dutz" <[email protected]>: Hi all, last night I had another idea what we should check before triggering a new release ... I was sort of wondering why we have to add an empty byte if the data is only one byte long. Then I thought ... could it be that the device is using a WORD padding? So it expects every part to be of an even number of bytes. If that was the case, if we read for example 3 bytes, we would get an additional fill byte too. Then we should verify and eventually fix this before pushing out a 0.3.1. Chris Am 11.02.19, 19:26 schrieb "Julian Feinauer" <[email protected]>: Hey all, I agree we should do this of 0.3 branch. Would be a good exercise with cherry picking and so. I suggest that I do the rm and Tim does all steps with me {we're working together}. Is this okay Tim or do you want to do everything on your own.? Julian Von meinem Mobiltelefon gesendet -------- Ursprüngliche Nachricht -------- Betreff: Re: Bugfix-Release 0.3.1 Von: Christofer Dutz An: [email protected] Cc: Hi Tim, I have no objections ... I guess you are in possession of a signed key? (I think us signing stuff at the Finka ... but if not, Julian could always sign your key) And having more people able to release is never a bad thing. If this bug is causing you to have problems in production, I agree and we should send a bugfix version (That would be released from the existing 0.3 branch). So the bugfix would have to be cherry-picked into that branch. I doubt it qualifies for a full release (0.4.0) as the only other significant change would have been my work on the dynamic driver. Chris Am 11.02.19, 16:13 schrieb "Tim Mitsch" <[email protected]>: Hallo everybody As we just released version 0.3, we found (and already fixed within develop-branch – thanks to Chris) a bug regarding exchange of One-Byte-long variables within S7, where sometimes a filling Byte hast o be used. This maybe leads to some strange behavior and it maybe be better to release a bugfix version where this bug is fixed. Julian (thanks for taking care about release of 0.3) as leader and supervisor and me (I have to learn it as well) would care about the release process for this bugfix release. What do you think? Best Tim
