array printing only shows the first few columns now

[ViralBShah]

Earlier, array printing used to show the first few and the last few columns. Now, it only shows the first few. However, in the case of rows, it still shows the first and last few rows. It would be nice to have the original behavior of showing the first and last few columns.

julia> rand(100,100) 100x100 Float64 Array: 0.9855081647166546 0.7290594996053321 0.4520218071241384 : 0.9459221906691384 0.6523605246578399 0.0390424896981334 0.9082303261548277 0.14640353894875724 0.978974074453369 0.9831170811607819 0.5633738106088249 0.39999133833394396 0.49178750990051645 0.8261447620770301 0.1373687161141457 0.19815512698323468 0.07321565789387718 0.6713381196638348 : 0.7485089268850265 0.3110337389266493 0.8127260070007878 0.09225990672270057 0.5164689845360908 0.41622752801604057 0.6463331342211742 0.631728035674942 0.8432739116222261 0.9589961098014961 0.759966809218134 0.8773805898884042 : 0.44189054473853373 0.7582624831117313 0.27365008249454736 0.3789754476891729 0.7282745467614227 0.26109809717894983 0.5755478924574304 0.744431037975464 0.8280690881815014 0.31616385733163543 0.49213861339515086 0.5900419739032776 0.16919930051323973 0.6495643537948732 0.07744277889228335 : 0.4150890974477517 0.17595839381354095 0.01932111769978051 0.45815810997355433 0.4925601322132367 0.2417257132734527 0.43765247512322825 0.9175191887051555 0.11772815227330469 0.27087472081338726 0.6977960270235686 0.5431813055965915

[ViralBShah]

Also, at some point we were printing only 7 or 8 digits after the decimal. Are things in flux due to the grisu integration?

[JeffBezanson]

Yes, it seems to be in flux. Another issue is that we have both showcompact and print_shortest, and that situation needs to be cleaned up somehow.

[StefanKarpinski]

I can try to fix this. Sorry the state of printing has been in such flux. The showcompact function is for showing things in arrays and other places where you want a compact but still nicely human-readable version of something. print_shortest is meant to give the tersest possible exact representation of a floating-point value, for printing lots of numeric data, e.g. in TSV or CSV files.