From: <pl...@pi...> - 2014-05-12 16:14:15
|
On 05/12/14 17:06, sfeam wrote: > > On Monday, 12 May 2014 08:42:55 AM pl...@pi... wrote: >> Hi, >> >> using gnuplot v5 built yesterday I find the fit "Deprecated syntax" >> warning breaks input. >> >> >> I linux primary buffer ( mouse 3 click to paste ) to copy data from a >> text as input to gnuplot's special filename '-' >> >> >> fit [0:fit_end] f_peak2(x) '-' using 1:2 via FWHM2,a >> #4 3.5 >> 10 3.47 # 3.5 >> 25 3.79 #4.43 >> 50 4.45 #6.78 >> #100 11.755 # 60.9+56.65 >> #100 11.21 >> 200 4.48 #21.96 >> e >> >> Instead of just printing the warning and plotting as it did with 4.6 it >> looses the rest of the input after the "fit" command and sits there with >> the data input prompt. > > I seriously doubt that this has anything to do with printing or > not printing a warning message. I have seen similar failure to > paste inline data from the mouse buffer even for simple plots > without any involvement of "fit". I don't know when or why > it occurs. If you can pin down a point where it changed, perhaps > by bisecting builds from old CVS, then maybe we can figure out > how to avoid this failure. > > Ethan > > Well nothing seems to want to play now :( last nights CVS , 4.6.5 and my old 4.5 cvs all produce this error now. G N U P L O T Version 4.6 patchlevel 5 last modified February 2014 Build System: Linux i686 G N U P L O T Version 4.5 patchlevel 0 last modified 2012-01-17 Build System: Linux i686 G N U P L O T Version 5.0 patchlevel alpha last modified 2014-05-10 However, I have found a work around. If I paste the fit command , then in a separate operation paste the data it takes it. BTW when it fails I need to enter two "e" lines to clear it. Whatever is on the first line gets ignored and a new prompt for data. fit [fit_min:fit_end] f2(x) '-' using 1:2 via m2,c2 fit: Deprecated syntax. Consider using the 'noerror' option, see `help fit`. input data ('e' ends) > e input data ('e' ends) > e Read 1 points Skipped 1 points outside range [x=50:200] No data to fit You are correct , nothing to do with the error message, adding "noerror" to comply does not get rid of the problem. In passing the new keyword "noerror" is not clear and does not scan. It sounds like a option to suppress errors or error reporting. noerrordata would be clearer. If something like that could used instead, anyone using the relatively new feature as "noerror" would still be OK because it would be an unambiguous abbreviation. my2c. Why was this needed, can't it just default to that anyway? yet another verbose option to remember before fit will even work at all. Peter. |