bug-cvs
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: issues in case folding as a result of a OS/2 to UNIX migration


From: Shaun Tancheff
Subject: RE: issues in case folding as a result of a OS/2 to UNIX migration
Date: Mon, 29 Sep 2003 16:38:23 -0500

Ahh. So my replacement has arrived.

> -----Original Message-----
> From: bug-cvs-bounces+shaun.tancheff=csausa.com@gnu.org 
> [mailto:bug-cvs-bounces+shaun.tancheff=csausa.com@gnu.org] On 
> Behalf Of Mesfin, Alex
> Sent: Monday, September 29, 2003 1:12 PM
> To: bug-cvs@gnu.org
> Subject: issues in case folding as a result of a OS/2 to UNIX 
> migration 
> 
> 
> We have an OS/2 running CVS 1.11.2.2 server, we are in the 
> process of migrating to sun box Solaris 2.8/9. 

> The problem are as follows:
> *     In OS/2 file and directory naming a case is irrelevant, 
> i.e. a user can add or checkout a file or directory with out 
> any concern to the case used in naming.
>               Ex.
>                       cvs add file.c  will succeed
>                       cvs commit File.c    will succeed
> 
>                       cvs add file.C  will fail 
>                       cvs co FILE.C         will succeed 
> 
>               As those example above indicated,  OS/2 is a 
> case stupid (Case Challenged). 
> 
>       My task is:
> *     prevent people from adding file in the new Unix box, 
> that only differ in case already in repository.

Hack the add.c code to do a directory lookup and compare with
strcasecmp.

> *     silently update files if they exist with a different 
> case in repository.

In what cases? 
If the current repository file is file.c,v should:
  cvs commit -m "SCR #99999." FILE.c re-fix the case to FILE.c ?
  cvs checkout, update, tag, diff, edit, etc.

> I am looking in ways to tackle this problem by 
> modifying/adding codes in to the cvs code. I'm asking for any 
> person who can share his/her insights on how the CVS process 
> flow works internally. Ex. 
>       when a client calls server obviously the main is the 
> entry point. Server from main goes to the server (server.c). 
> 
>       How does client pass information to server and where in 
> server this information is processed.
> 
>       I know that people who worked on this have an easy time 
> navigating the source tree, I would be very thankful if they 
> could share their insights.

You really shouldn't be concerned about the 'difference' between the 
server and the client (the primary differenece in the code path is
with regard to communication in server.c and client.c).

If you can follow my AdmUsageLog hack you pretty much grok cvs
client/server.

>       
> 
> Eskinder A. Mesfin CRM - AM Tools
> Sr. SW Engineer
> Medtronic Inc.
> (763) 514-8267
> eskinder.a.mesfin@medtronic.com
> 
>  <<Mesfin, Alex.vcf>> 
> 

-- 
http://shaun.tancheff.com/              /?/   -_-   /
  :-) shaun ;-] shaun.tancheff@csausa.com







reply via email to

[Prev in Thread] Current Thread [Next in Thread]