I guess the idea of restoring standard definitions before running a lab can be even more dangerous, as it might overwrite a user's work if he happened to define something with a name used in the standard library.
Maybe the path of the locale the user is in when starting the lab can be appended to the one of the theorized "lab" locale, so that there is at least read-only access to things users defined before , without having to use locatives ... 2015-05-26 21:48 GMT+02:00 chris burke <[email protected]>: > It is a nice idea, but labs were intended to be run in a normal session. > While stepping through a lab, the user would be able to experiment with the > definitions and run any other code. If the labs were run in a "lab" locale, > this would not work quite so well. > > I suppose an alternative is having some way of restoring all standard > library definitions, and doing this automatically before a lab is run. > > On 26 May 2015 at 12:29, Jan-Pieter Jacobs <[email protected]> > wrote: > > > Yes I was. > > > > The errors I was seeing already have been fixed by Henry Rich. > > The breakage of the tabula addon was partly worked around by Bill Lam. > > > > Conceptually, it would be nice to have all labs run in a "lab" locale, > > which is cleared on entering a new lab. That way users that want to copy > > stuff from a lab can still do so, there's no conflicts with user defined > > values and lab's don't have to care about redefining already existing > > stuff. > > But probably I'm missing useful behavior which would break this way. > > > > Just my 2 cents. > > > > Jan-Pieter > > > > 2015-05-26 20:03 GMT+02:00 chris burke <[email protected]>: > > > > > Running through the labs in order they show up gave me some problems: > > > > > > 0) Book of numbers: (4) 12 of 23 : domain error: on > > > 1) CAL : 3 of 11 : z =:tabengine 'CTAB' : length error SP1: n (x,.SP) > > ,. y > > > > > > After doing a "clear '' " both worked fine. So I assume that these were > > due > > > to > > > 0) other labs overwriting "standard" words (like "on") with something > > else > > > 1) labs not verifying what they use is what they expect it to be. > > > > > > Maybe it would not be a bad idea to run labs in a private locale, > > > which is initialized in a certain way? This way these conflicts (and > > > conflicts with user defined things) might be avoided. > > > > > > --------- > > > > > > This has been a problem in the past, and I had hoped we had eliminated > > all > > > the conflicts. In theory a lab shouldn't change standard library > > > definitions. > > > > > > It would be too much to expect that a lab should verify standard > library > > > definitions, but anything else should be defined in the lab itself. > > > > > > Unfortunately, I couldn't reproduce either problem by running through > the > > > labs in order. Are you running the latest labs and J803? > > > > > > > > > On 4 May 2015 at 01:19, Jan-Pieter Jacobs <[email protected]> > > > wrote: > > > > > > > Hi everyone, > > > > > > > > I found some bugs I'd want to inform about: > > > > * Infinite popups in plot > > > > * comments syntax highlighting broken > > > > * Tabula add-on broken > > > > * Labs break each other. > > > > > > > > I just found another instance of an error which keeps popping up > > > > infinitely when using plot. > > > > A try to debug the error with the JQt debugger ended in a crash. > > > > > > > > A minimal example exhibiting this behavior is: > > > > > > > > load 'plot' > > > > opts =: 'xlabel myxlabel; ylabel myylabel' > > > > opts plot ;/~ i.10 > > > > > > > > Probably it is me doing something wrong, but as this makes me > forcibly > > > > shut down J every time I hit this error, it's pretty annoying to hunt > > > > down my error. > > > > ------------------------ > > > > Another thing I discovered when toying around was that syntax > > > > highlighting is slightly broken for comments in these cases: > > > > > > > > * a comment contains 0 : 0 : the entire comment (and following code > > > > till the next lone opening parenthesis) is regarded as a long string > > > > literal, and marked up as such. > > > > * if a word (probably any valid identifier) is directly followed by > an > > > > opening brace (even in comments) it's highlighted as a string. I > guess > > > > this is supposed to indicate functions in traditionally mathematical > > > > notation like f(x), but it should not happen in comments. > > > > ------------------------ > > > > Further, I've found the "math/tabula" add-on to be utterly broken, in > > > > that clicking any of the buttons pops up the error: > > > > > > > > error in: tab_g_mbldown > > > > > > > > domain error: chkgl2 > > > > glpixels X,Y,32,32,,convicon y > > > > ------------------------ > > > > Running through the labs in order they show up gave me some problems: > > > > > > > > 0) Book of numbers: (4) 12 of 23 : domain error: on > > > > 1) CAL : 3 of 11 : z =:tabengine 'CTAB' : length error SP1: n > (x,.SP) > > > ,. y > > > > > > > > After doing a "clear '' " both worked fine. So I assume that these > were > > > > due to > > > > 0) other labs overwriting "standard" words (like "on") with something > > > else > > > > 1) labs not verifying what they use is what they expect it to be. > > > > > > > > Maybe it would not be a bad idea to run labs in a private locale, > > > > which is initialized in a certain way? This way these conflicts (and > > > > conflicts with user defined things) might be avoided. > > > > > > > > Best regards, > > > > > > > > Jan-Pieter > > > > > ---------------------------------------------------------------------- > > > > For information about J forums see > http://www.jsoftware.com/forums.htm > > > > > > > ---------------------------------------------------------------------- > > > For information about J forums see http://www.jsoftware.com/forums.htm > > > > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
