hi, I had the same latency problem when using kite, when I type import datetime as dt : it's pretty much instant, but when I type dt. I have to wait around 15 s before it shows results, idk if it's because of my computer or it's a common issue.
Le mercredi 16 juin 2021 à 19:31:46 UTC+1, [email protected] a écrit : > Hi Vicent > > Thanks. > I just upgraded my Anaconda installation to 5.0.4 using: > > conda install --channel conda-forge spyder=5.0.4 > > Then I tried carrying out your fix. > > It has helped slightly, now it takes about 5 seconds for: > np. > to show the autocompletion list opposed to about 30 s. > > > The line > > import numpy a > > Now autocompletes to: > > import numpy as np > > [image: 62a.png] > > But > > np. > > still takes about 5 s to display. An improvement of the 30 s lag from > before but still slower than it should. > [image: 62a2.png] > > In comparison: > > import datetime as dt > dt. > > is pretty much instant: > > [image: 64.png] > > Kite gives the most commonly used methods/attributes at the top followed > by everything else ordered alphabetically. > I've noticed from time to time, Kites most commonly used recommendations > are a bit ropey in general. > > However for a numpy array in particular it is not exactly giving correct > recommendations for an array instance. > In the below, x is a numpy array instance. > > In this case it gives str methods which do not apply to the object x: > [image: 62.png] > > The list of methods/attributes below this are the correct ones belonging > to the array: > > [image: 63.png] > > Seems there is some confusion to its recommendations. > > > > On Tuesday, June 15, 2021 at 4:54:58 PM UTC+1 [email protected] wrote: > >> Hey, I had the same latency problem and found a way to solve it. For >> information, I use Spyder 5.0.0. >> >> 1) Open Kite. >> 2) Click on the parameter button in the bottom right corner. >> >> [image: googlehelp1.png] >> >> >> 3) Go to Plugins. >> >> [image: googlehelp2.png] >> >> 4) Within the category "AVAILABLE EDITOR PLUGINS", install the Spyder IDE >> one. >> >> >> [image: googlehelp3.png] >> >> I did this and autocompletion worked again. >> >> I hope this will help you. >> >> Best regards. >> Le dimanche 2 mai 2021 à 22:05:18 UTC+3, [email protected] a écrit : >> >>> It is still extremely slow without Kite installed. I guess it is related >>> to this issue de to be resolved in Spyder 5.0.3. >>> >>> Completions extremely slow · Issue #15042 · spyder-ide/spyder · GitHub >>> <https://github.com/spyder-ide/spyder/issues/15042> >>> >>> On Saturday, May 1, 2021 at 10:49:58 AM UTC+1 yfprojects wrote: >>> >>>> Hi, >>>> the *ipython* console/ *qtconsole* never uses *Kite*. As far as I know >>>> Kite only supplies extra completions. The display of the doc strings and >>>> the method/function signature is provided by another linting tool - >>>> probably *jedi*. The whole linting and completion functionality is >>>> bundled in a *python language server *(pyls). The server should >>>> provide completion options along with kite's completion. If kite is too >>>> slow on your PC, try to disable it. >>>> >>>> Kind regards, >>>> yfprojects >>>> >>>> [email protected] schrieb am Freitag, 30. April 2021 um 12:04:29 UTC: >>>> >>>>> I have been using both JupyterLab 3.0.14 and Spyder 5.0.1 and I >>>>> noticed that the experience with Spyder 5.0.1 is a bit frustrating mainly >>>>> because of the code auto-completion. This may be more of a Kite than a >>>>> Spyder issue... >>>>> >>>>> In JupyterLab, the code completion does not automatically display but >>>>> will display when tab is pressed. For example if I import numpy as np and >>>>> type in: >>>>> >>>>> np.↹ >>>>> >>>>> I get a list of all the statements/functions/classes I can pull from >>>>> the numpy library. >>>>> >>>>> [image: 1 jupyterlab.png] >>>>> >>>>> Then if I type in a function or class and press shift and tab I get >>>>> details about the input arguments. e.g. >>>>> >>>>> np.array⇧↹ >>>>> >>>>> [image: 2 jupyterlab.png] >>>>> >>>>> This works pretty well for numpy, matplotlib, pandas... >>>>> >>>>> [image: 3 jupyterlab.png] >>>>> >>>>> Moving onto Spyder 5.0.1 with Kite... >>>>> >>>>> If I import numpy as np and then type np. in the script nothing >>>>> displays. >>>>> >>>>> [image: 4 spyder.png] >>>>> >>>>> I need to run the script with the solo command import numpy as np >>>>> before using any autocompletion in the script with numpy (partially) >>>>> works. >>>>> >>>>> [image: 5 spyder.png] >>>>> >>>>> After about a minute latency the following displays... >>>>> >>>>> [image: 6 spyder.png] >>>>> >>>>> [image: kite.PNG] >>>>> >>>>> I can then select from the list the class or function that I want: >>>>> >>>>> [image: 7 spyder.png] >>>>> >>>>> Next Spyder has both the Script Editor and the Console. If I now go to >>>>> the Console and type in np. nothing happens. If I type in np.↹ then the >>>>> following autocompletion shows. I think this is Jedi opposed to Kite... >>>>> >>>>> [image: 9 spyder.png] >>>>> >>>>> Now if I type in a class/function with open parenthesis. This seems to >>>>> work with Kite both in the script editor and console. >>>>> >>>>> [image: 10 spyder.png] >>>>> >>>>> The performance is slightly better for matplotlib. There is about a 30 >>>>> s delay with plt. opposed to about 60 s with np. Either way the delay is >>>>> still too long hindering productivity. >>>>> >>>>> [image: 11 spyder.png] >>>>> >>>>> However once again typing plt. into the console does not generate a >>>>> Kite autocompletion. >>>>> >>>>> [image: 12 spyder.png] >>>>> >>>>> Typing in plt.↹ generates a Jedi autocompletion. >>>>> >>>>> [image: 13 spyder.png] >>>>> >>>>> >>>>> If I type in "import numpy as np" as an earlier line in the script. >>>>> Should np. not generate autocompletion both later on in the script and in >>>>> the console using Kite without the latency issue (without previously >>>>> running the script). >>>>> >>>>> Also can the behaviour of Kite be changed so it doesn't pop up >>>>> automatically and only pops up (no delay) if requested? i.e. uses the >>>>> same >>>>> shift and tab shift keyboard combination as JupyterLab. >>>>> np.↹ >>>>> np.array⇧↹ >>>>> i.e. exhibits similar behaviour (to Jedi) and the code competition in >>>>> JupyterLab. >>>>> Alternatively, if that is not possible to do with Kite is there a way >>>>> to use the code completion from JupyterLab within Spyder instead of Kite? >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> -- You received this message because you are subscribed to the Google Groups "spyder" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/spyderlib/cff06a05-232c-44aa-b9c4-4ed61bf7e241n%40googlegroups.com.
