Very slow boot time on Mac app

[myeomans]

Hi, I switched from windows to Mac recently and noticed that Spyder takes a very long time to start up when I try to open the program. I'm talking, like, 30min to maybe an hour or two sometimes. I still use it because it's a phenomenal program, but the long boot time is really annoying. It didn't happen on my old windows machine, so I don't know what the issue is. Any idea what's up? If there's a simple fix I'd love to know. I hate having to wait to use the console.

[ccordoba12]

Questions:

1. What Spyder version are you using?
2. What Python version?
3. How did you install Spyder?

We have a menu entry under the Help menu called Report a bug that will give us all this information (and more) when you report a bug here (since Spyder 2.3.3), so please use it next time that you have troubles.

[myeomans]

Thanks for your reply, I submitted my issue using the "report a bug"
feature instead, that will be more helpful I think.

Best,
Mike

On Sun, Mar 29, 2015 at 2:28 PM, Carlos Cordoba notifications@github.com
wrote:

Questions:

1. What Spyder version are you using?
2. What Python version?
3. How did you install Spyder?

We have a menu entry under the Help menu called Report a bug that will
give us all this information (and more) when you report a bug here (since
Spyder 2.3.3), so please use it next time that you have troubles.

—
Reply to this email directly or view it on GitHub
#2284 (comment).

[Nodd]

I can't find the new issue ?

Instead of creating a new one, you can just open the page with spyder's menu and copy-paste the content in this one. Either way is fine !

[Nodd]

It's a duplicate of #2171, right ?

[myeomans]

Not a duplicate, though #2171 sounds very similar to my issue. I checked and can confirm that the % CPU does go to 100% in the activity monitor while I'm waiting (though my computer is multicore so the overall percentage in use is still modest).

The people in the other thread don't have to wait nearly as long to load as I am (10 minutes would be a dream at this point!) Mine is taking more like an hour or two. My splash screen takes a few minutes once it is up, but that doesn't bother me much. During the long wait before the splash, the program is giving me no indication that it's doing anything... Also I am running Yosemite 10.10.2, not Mavericks.

Once/if spyder boots today I will paste my report in this thread, as requested.

[myeomans]

Okay, here's the bug report info:
Spyder Version: 2.3.2
Python Version: 2.7.8
Qt Version : 4.8.6, PyQt4 (API v2) 4.11.1 on Darwin
pyflakes >=0.5.0: 0.8.1 (OK)
pep8 >=0.6 : 1.5.7 (OK)
IPython >=0.13 : 2.3.1 (OK)
zmq >=2.1.11 : 14.4.1 (OK)
pygments >=1.6 : 2.0.1 (OK)
pandas >=0.13.1 : 0.15.2 (OK)
sphinx >=0.6.6 : 1.2.3 (OK)
psutil >=0.3 : 2.1.3 (OK)
rope >=0.9.2 : 0.10.2 (OK)
matplotlib >=1.0: 1.4.2 (OK)
sympy >=0.7.0 : 0.7.6 (OK)
pylint >=0.25 : 1.4.0 (OK)

[ccordoba12]

@myeomans, please update to Spyder 2.3.4. I think I fixed this problem in 2.3.3.

[myeomans]

Yeah, the update did it! Thanks so much!

[ccordoba12]

@myeomans, glad to know things are working fine for you again! :-)

[leonsuetfeld]

Unfortunately I don't think 2.3.4 fixes it.

I'm using a Retina MBP, OSX 10.10.3 and python 2.7. When I used Spyder 2.3.2 it would always take very varying amounts of time, say between 10 and 120 minutes, to start up Spyder.

I then switched to 2.3.4 and the problem SEEMED fixed, so it always started up within a few seconds. However, after a crash it all went bad again, now showing the exact same behavior as 2.3.2 did before. I think the may have been caused by me trying to open the variable explorer while the code was running - it seems to be very sensitive about that. I'm not sure though, don't remember exactly.

Bottom line is, something in 2.3.4 fixed it at first but now the problem is back.

[myeomans]

Agreed with leonsuetfeld, the problem is back... though I didn't have any "crash" moment that changed things. I don't ever use the variable explorer, and rarely run a lot of code at once (i mainly use the interpreter to test out small things before I load them onto a python server). I'm not ruling out the possibility that I opened something that triggered the problem by accident.

[ccordoba12]

Hi,

I need a reproducible test to fix this problem because I'm not seen it myself :-) For now you can try to run in a terminal:

spyder --reset

or

rm -Rf ~/.spyder2

to see if that fixes the problem.

@leonsuetfeld: what variable were you seeing when using the variable explorer?

[ccordoba12]

@blink1073, are you seeing this problem in your machine?

[blink1073]

I am not. 

—
Sent from Mailbox

On Mon, May 25, 2015 at 1:27 PM, Carlos Cordoba notifications@github.com
wrote:

@blink1073, are you seeing this problem in your machine?

Reply to this email directly or view it on GitHub:
#2284 (comment)

[leonsuetfeld]

What folder would I have to be in to execute the commands you suggest?

I think just re-installing the dmg file of spyder 2.3.4. may have done the trick. At least start-up time is back to what it should be (<10s), for now. So there probably is something that can go wrong after some time (or a spyder crash possibly), that can be fixed by re-installing spyder 2.3.4.

About the variable explorer: If you're referring to the time it crashed after which the problem may have occured for the first time: It didn't get to that point when it showed any variables. It crashed right after I requested the VE, since it wasn't happy about requests while executing code.

[ccordoba12]

@leonsuetfeld, if you're using our dmg, you can't run spyder --reset (you'll be able to since Spyder 3.0 :-), but you can run rm -Rf ~/.spyder2 on any folder (using Terminal.app of course).

Glad to hear that your problem was fixed :-)

[zach-morris]

I put the same comment in a different thread of a similar problem, but I had this same exact issue. After a random crash of the application, it would take up to 10 minutes to start up. After looking at the Package Contents, I noticed that a file name 'pep8' was getting very large (over 1GB). After I deleted that file, my slow startup issue went away completely.

[myeomans]

This problem still isnt' fixed. The program can run after installing for a month or two and then slows down dramatically at load. If I have to keep reinstalling the whole thing each time it starts freezing at the program load, that makes it very difficult to rely on long-term.

I found a few pep8 files in the directory but none of them were very big (all <100kB) so I don't think that's the issue in my case.

[Cognitively]

Same here. Using 2.3.7 on OS X 10.10.5. I had to remove spyder and reinstall it. The older 2.3.7 Spyder.app was ~900MB, the freshly installed 2.3.7 is ~500MB

[zach-morris]

Just installed this on a new mac (10.11.2) and it looks like the same problem exists. I checked the app contents and found that the files pep8 and pylint continually are getting larger and larger in size.

Fresh install. Before launching Spyder, the file size was as follows:
pep8: 3kb
pylint: 3kb

Start spyder, the filesize jumps:
pep8: 5kb
pylint: 5kb

Restart the application. File sizes jump again:
pep8: 10kb
pylint: 10kb

After trying this a few times, I see these files are doubling in size every time I restart the application. I opened the files, and it looks like the file is just doubling the newlines/return carriages every time the application launches. Seems like it should be a simple fix.

[ccordoba12]

@zach-morris, thanks for the update. I think I know what's happening, but I have to do some experiments to confirm it.

[ionoDORK3]

This is a significant problem for our research group. We must use Python 3.4 on MacOS X, which is part of the Spyder 2.3.7 release. As mentioned above, the pep8 and pylint files explode in size each time the Spyder IDE runs. We realize that this is an older version, but we would really appreciate it if this problem could be investigated and solved.

The release is:
URL: https://github.com/spyder-ide/spyder/releases
FILE: spyder-2.3.7-py3.4.dmg

Thanks.

[ccordoba12]

@ionoDORK3, I'm sorry but I'm unable to produce more DMG's with Python 3.4 because it's not supported anymore in Homebrew.

You could Anaconda instead :-)

[myeomans]

Is there any version that doesn't have this problem? I keep a DMG handy
to re-install spyder every few days when I'm using it.. This is about as
ugly a solution as it gets.

On Wed, Mar 2, 2016 at 11:19 AM, Carlos Cordoba notifications@github.com
wrote:

@ionoDORK3 https://github.com/ionoDORK3, I'm sorry but I'm unable to
produce more DMG's with Python 3.4 because it's not supported anymore in
Homebrew.

You could Anaconda instead :-)

—
Reply to this email directly or view it on GitHub
#2284 (comment).

[ionoDORK3]

1. TO: ccordoba12
   We are stuck with this version of Spyder/Python.
2. TO: myeomans
   Our only workaround (so far) is to copy small versions of the pylint and pep8 files into the Resources folder now and then. When those files are small, Spyder will come up quickly. I bet using the DMG does the same thing. Both solutions are not cool...

[myeomans]

Right, exactly. our solutions are surely doing the same thing to provide
the temporary fix.

I don't want to be a complainer because Spyder really is a good programming
tool! But this one bug is what keeps me from recommending it to colleagues
who want to start using python themselves.

On Wed, Mar 2, 2016 at 2:22 PM, ionoDORK3 notifications@github.com wrote:

TO: ccordoba12
We are stuck with this version of Spyder/Python.
2.

TO: myeomans
Our only workaround (so far) is to copy small versions of the pylint
and pep8 files into the Resources folder now and then. When those
files are small, Spyder will come up quickly. I bet using the DMG does the
same thing. Both solutions are not cool...

—
Reply to this email directly or view it on GitHub
#2284 (comment).

[SjoerdOptLand]

Here, just deleting pep8, pylint and pylint.bak seems to unblock Spyder (it recreates them). Is deleting these files harmful in any way?

[ccordoba12]

You won't be able to use our integration with those libraries, but there isn't harm in removing them because they are not core functionality for Spyder :-)

[hdj]

I also encounter the same problem. The size of pep8, pep8.bak and pylint are getting larger and larger in size and it takes a long long time to open spyder.
By following the solution mentioned above, I delete the pep8, pep8.bak and pylint.
The Spyder becomes normal again.

I install spyder by using spyder-2.3.8-py2.7.dmg. My OS version is 10.11.5
Hope this information can help the authors to improve the spyder.

[ptoche]

Great tip SjoerdOptLand!

Problem: slow startup or spyder gets stuck starting up.
Solution:

1. Go to spyder.app
2. Show package content
3. delete pep8, pep8.bak, pylint and pylint.bak (these files are greater than 100MB each)

[ccordoba12]

I finally found the problem with this issue. It'll be fixed when I release new DMGs along with Spyder 3.0.1 :-)

[goanpeca]

What was it ?

[ccordoba12]

This function:

https://github.com/spyder-ide/spyder/blob/master/create_app.py#L173

which is added to the the file that boots our Mac app, was adding new, empty lines to its pep8 and pylint scripts every time Spyder was started. And that was making those script to increase in size and become really big files in very short time.

[goanpeca]

👍