|Title||"p4 -G client -o" sometimes fails|
|Assigned user||Gareth Rees|
|Description||"p4 -G -u username -c clientname client -o" sometimes fails to write output that Python's marshal.load can read.|
|Analysis||This problem is present in the Perforce 2000.1 release.|
I explored the envelope of the problem and it looks like it occurs when the length of the clientname argument is 8 characters.
It is essential that this problem is fixed, because otherwise the replication could fail mysteriously when creating or examining Perforce clients. This would affect requirements 1 and 63.
Chris Seiwald discusses the problem and proposes a solution in [Seiwald 2000-09-28a] and [Seiwald 2000-09-28b].
GDR tested the new Perforce client P4/NTX86/2000.1/17595 and changed the replicator so that it always writes to and reads from the Perforce process using binary streams. It works fine with 8-character client names.
For the alpha release, we'll need to use Chris' supplied fixed client on the machine with the replicator, but not elsewhere. RB 2000-10-06
|Evidence||[GDR 2000-09-26] "'p4 -G client -o' fails" (e-mail message); Gareth Rees; 2000-09-26 15:34:27; <URL: |
[Seiwald 2000-09-28a] "Re: 'p4 -G client -o' fails" (e-mail message); Chris Seiwald; 2000-09-28 09:08:49 GMT; <URL:
[Seiwald 2000-09-28b] "Re: 'p4 -G client -o' fails" (e-mail message); Chris Seiwald; 2000-09-28 10:14:31 GMT; <URL:
[Seiwald 2000-09-28c] "Re: Line endings in 'p4 -G job -i'?" (e-mail message); Chris Seiwald; 2000-09-28 19:49:45 GMT; <URL:
|Created by||Gareth Rees|
|Created on||2000-09-26 16:43:45|
|Last modified by||Gareth Rees|
|Last modified on||2010-10-06 21:19:43|
|History||2000-09-26 GDR Created.|
2000-09-28 GDR Referenced [Seiwald 2000-09-28a] and [Seiwald 2000-09-28b].
2000-09-29 GDR Referenced [Seiwald 2000-09-28c].
2000-10-06 RB Added note about alpha release workaround.
2000-11-28 RB Tidied up. Closed (see evidence).
|5529||closed||2000-12-06 12:13:09||Richard Brooksby||Changing the UG to reflect the "p4 submit -s" behaviour now that Perforce have fixed it.
Changing the AG to insist on a particular changelevel of the server that includes that fix, in case anyone has downloaded an earlier Perforce 2000.2 beta and thinks it will work.
|2585||open||2000-09-28 17:41:00||Gareth Rees||The Python interface to Perforce writes to and reads from the Perforce process using binary streams. See job000010.|
|2576||open||2000-09-28 14:47:50||Gareth Rees||Added Perforce client NTX86/2000.1/17595 to imported materials.|