testAll errors

[Bhushan Dhamale]

I attempted to create a log of the command window output for the testAll command by using diary command, but for some reason that eludes me, the log file refuses to record the output beyond a limit (the file size always gets stuck at (96.9 kB). I have included the testAll summary in the attachments.
Test all throws several errors, a few of which I have included in the attachments as well.
Could someone please suggest me a fix to this? I need to fix this urgently!

[Laurent Heirendt]

Hi,

May you attach the report generated by >> generateSystemConfigReport ?

Best,

Laurent

[Thomas Pfau]

Hi,

From the error messages it looks like you neither have the parallel toolbox nor the bioinformatics toolbox (and probably other things are missing as well.)

I'll have a look at some parts of the code (as I expected, that at least the missing parallel toolbox just increases computational time and does not lead to so many failing tests).

Best

Thomas

--

---
You received this message because you are subscribed to the Google Groups "COBRA Toolbox" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cobra-toolbo...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

-- 
Université du Luxembourg
Faculté des Sciences, de la Technologie et de la Communication
Campus Belval, Biotech II 115
6 avenue du Swing
L-4367 Belvaux
Tel: (+352) 46 66 44 5309
Email: thoma...@uni.lu

[Bhushan Dhamale]

I have attached the report.

[Bhushan Dhamale]

Hi Thomas.
Is Bioinformatics Toolbox absolutely essential for the proper functioning of COBRA Toolbox?
Also, a few of the errors (e.g. testFVA, and testSampleCbModel) seem to be resulting due to the lack of Parallel Computing Toolbox.
Other than installing the toolboxes, what might be done to fix these errors?

[Bhushan Dhamale]

Hello Laurent and Thomas,
I installed the parallel processing toolbox, bioinformatics toolbox. statistics and machine learning toolbox, and optimization toolbox. It seems to have fixed 7 of the 17 errors that I was getting in the failure summary. But, I'd still very much prefer for it to work without the additional toolboxes as I'm skeptical about our lab being able to afford to purchase all the toolboxes.
Do you have any suggestions that might help me sort this out?

P.S.: I am also attaching the new error summary for your reference.

[Thomas Pfau]

Hi,

There are some things that we won't code ourselves, but anyone is free to provide an implementation and add it to the toolbox. I'm currently looking through the functions/test to see which could be run without toolbox code and update them if an easy fix is possible.

About your additional errors:

Without having the actual error messages, it is hard to tell what might be going wrong. Could you run those tests individually and report the errors?

Best

Thomas

[Bhushan Dhamale]

Hello Thomas,
I sequentially uninstalled the toolboxes and created a failure summary for each instance. In the file, I have highlighted the errors that occur after the uninstallation of each toolbox. Also, I have organized the error messages in the same order.
Hope this helps.

Best,
Bhushan

[Thomas Pfau]

Hi Bhushan,

First of all: Thank you for the effort. It helped in determining at least some issues.
Now, actually only a part of these errors is due to misisng toolboxes. A signioficant part stems from this seemingly being an outdated version of the toolbox.

E.g. the model folder changed some time ago, and it seems, that, while the new folder was created, the tests on your system were not updated.
I would recommend to update the toolbox either by using updateCobraToolbox or directly via git commands if you are comfortable with those.

the git commands would be:

> git checkout master

This ensures, that you are working on the master branch.

> git fetch

This downloads all updates from git.

> git reset --hard

This remove all changes you have done locally (and it is not reversible, so maybe you want to create a copy if you want to keep your changes)

> git pull

This updates your version to the latest version from the git repository

> git submodule init

> git submodule update

if the latter commands fail, delete the test/models directory manually and repeat them. This is likely due to a change in how models are provided with the toolbox.

If you don't want to go the full git route, you can simply delete your current installation and create a new clone.

Once you updated your toolboxes a lot of tests that currently fail should succeed.
However, success of a test does not necessarily indicate that the function works on your system.

We are currently working on getting the tests to report the problems.

Best

Thomas

[Bhushan Dhamale]

Hello Thomas,
Thank you for the prompt reply.
I updated the toolbox using the git commands. After the update, the number of errors have been slashed down from 53 errors to 20 errors.
I am attaching the failure summary along with the errors categorized under their toolbox of origin (more precisely, lack thereof).

Thanking you,
Bhushan

[Thomas Pfau]

Hi Bushan,

This sounds about right.
I'm currently working on reducing the number of toolbox dependencies but there will be some functions which will not be able to run without the toolboxes installed.
Especially for the parallel toolbox I think the only function from that category that does nto run without the parallel toolbox is fastFVA.

Best

Thomas

[Bhushan Dhamale]

Hi Thomas,
Could you please give me some pointers as to how I could use the 'Optimization Toolbox' and 'Parallel Computing Toolbox' utilizing functions without installing the toolboxes. The dependent functions I previously mentioned under these toolboxes happen to be of immediate relevance to me.

Best,
Bhushan

[Thomas Pfau]

Hi Bhushan,

For most of the parallel toolbox instances (except fastFVA):
Just use them. The problem here is, that the tests call functionality of the parallel toolbox and thus fail (this will be fixed in the near future), and thus the tests fail. The functions however mostly work fine without having the parallel toolbox. There is one bug in FVA for which a pull request with a fix exists.

As for the Optimisation toolbox:
If you want to use Non Linear things, you will need some non linear solver supported by the toolbox, and those are currently tomlabs snopt optimizer and matlabs own optimizer which is supplied with the Optimization toolbox.

If you fear that solveCobraLP wont work: Ignore that. This is due to setting the matlab solver not yet checking whether the optimization toolbox is installed. So as long as you have any other LP solver, you are fine.

Best

Thomas