This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

classification
Title: test_winreg failing, Win98SE
Type: Stage:
Components: Windows Versions: Python 2.4
process
Status: closed Resolution: fixed
Dependencies: Superseder:
Assigned To: gvanrossum Nosy List: anthonybaxter, gvanrossum, tim.peters
Priority: normal Keywords:

Created on 2003-12-01 20:01 by tim.peters, last changed 2022-04-11 14:56 by admin. This issue is now closed.

Messages (9)
msg19210 - (view) Author: Tim Peters (tim.peters) * (Python committer) Date: 2003-12-01 20:01
Under current CVS, on Win98SE:

C:\Code\python\PCbuild>python ../lib/test/test_winreg.p
y
Traceback (most recent call last):
  File "../lib/test/test_winreg.py", line 135, in ?
    TestAll(HKEY_CURRENT_USER)
  File "../lib/test/test_winreg.py", line 130, in TestAll
    WriteTestData(root_key)
  File "../lib/test/test_winreg.py", line 37, in 
WriteTestData
    SetValueEx(sub_key, value_name, 0, value_type, 
value_data)
WindowsError: [Errno 87] The parameter is incorrect

C:\Code\python\PCbuild>

That's all I know now.
msg19211 - (view) Author: Guido van Rossum (gvanrossum) * (Python committer) Date: 2003-12-01 20:14
Logged In: YES 
user_id=6380

I bet the Big String and/or Big Binary test values that I
added to the test_data table are too long for Win98. I can
only test this on Win98. Can you experiment and find out
what the largest value is that works on Win98? (The
alternative would be to write code that somehow tries to
*detect* the longest string that works, using bisection. I
don't know if these limits are documented somewhere.)
msg19212 - (view) Author: Tim Peters (tim.peters) * (Python committer) Date: 2003-12-01 20:28
Logged In: YES 
user_id=31435

Yes, when I reduced the size of Big String to 12, it stopped 
failing there and failed on Big Binary instead.  The docs don't 
promise that anything larger than 2048 will work ...

"x"*(2**14-1) works, "x"*2**14 does not, for Big String.

"x"*2**14 works, "x"*(2**14+1) does not, for Big Binary.

So 2**14 bytes is the max on my Win98SE, when counting 
the trailing zero byte in REG_SZ.

The limits are not documented (apart from the docs saying 
you shouldn't use any value exceeding 2**11 bytes).
msg19213 - (view) Author: Anthony Baxter (anthonybaxter) (Python triager) Date: 2003-12-02 00:23
Logged In: YES 
user_id=29957

Argh. Is this busted on the 2.3 branch as well?
msg19214 - (view) Author: Tim Peters (tim.peters) * (Python committer) Date: 2003-12-02 00:34
Logged In: YES 
user_id=31435

Anthony, yes, it is.
msg19215 - (view) Author: Guido van Rossum (gvanrossum) * (Python committer) Date: 2003-12-03 15:16
Logged In: YES 
user_id=6380

Sigh. I'll adjust the test to use 2**14-1 and 2**14,
respectively. But that no longer tests for the segfault I
was experiencing before... :-(

BTW, if I seem tardy responding to this, it's because I just
discovered that SF doesn't send me email when this item
changes (maybe it doesn't send me mail at all, I haven't
seen any SF mail in ages I think).
msg19216 - (view) Author: Guido van Rossum (gvanrossum) * (Python committer) Date: 2003-12-03 15:21
Logged In: YES 
user_id=6380

Sigh.  Somehow my SF email forwarding was set to
guido@zope.com. I've fixed it now... But who know how many
other reminders of SF items I've missed... (I wonder if this
was recently changed??? I sure remember getting at least
some SF mail recently, and I definitely get SF mail from the
IDLEfork tracker; I've only got one SF account.)
msg19217 - (view) Author: Guido van Rossum (gvanrossum) * (Python committer) Date: 2003-12-03 15:27
Logged In: YES 
user_id=6380

Fixed in CVS, for 2.4 and 2.3.  Let's hope this is it. (Tim,
please confirm.)
msg19218 - (view) Author: Tim Peters (tim.peters) * (Python committer) Date: 2003-12-04 07:09
Logged In: YES 
user_id=31435

Fix confirmed under 98SE on trunk and on 2.3 maint.  
Thanks!  Closed the bug report.
History
Date User Action Args
2022-04-11 14:56:01adminsetgithub: 39642
2003-12-01 20:01:40tim.peterscreate