So, as the title says this is weird. I can use tkcvs in a Git repo just fine. But when I try to use it in a CVS repo:
$tkcvscvsxdiffErrorinstartupscript:badoption"close":mustbecgetorconfigurewhileexecuting"$lc close invoke"(procedure"::cvs_branchlog::0::cvs_sort_it_all_out"line29)invokedfromwithin"[namespace current]::cvs_sort_it_all_out"(procedure"::cvs_branchlog::0::reloadLog"line80)invokedfromwithin"[namespace current]::reloadLog"(innamespaceeval"::cvs_branchlog::0"scriptline481)invokedfromwithin"namespace eval $my_idx {setmy_idx[uplevel{concat$my_idx}]setfilename[uplevel{concat$filename}]sethow[uplevel{concat$h..."(procedure"::cvs_branchlog::new"line6)invokedfromwithin"::cvs_branchlog::new "CVS,loc" "$file""(procedure"cvs_branches"line14)invokedfromwithin"cvs_branches [list $lcfile]"invokedfromwithin"if {[string match {mod*} $cvscfg(startwindow)]} {wmwithdraw.# If we're in a version-controlled directory, open that repositoryif{$insvn}{..."(file"/discover/swdev/mathomp4/local-SLES12/bin/tkcvs"line417)
I'm not sure how to decode that... Any help?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
That was interesting. The source code has been on the tigris.org collabnet ever since I made the Subversion module. Tigris closed down, and it took me all this time to get a dump so I could bring it back to Sourceforge.
Anyway, I fixed the proximate cause of the crash, which was this, but there was before that some reason why your revision-log file couldn't be processed. If it isn't proprietary, you could find the ,v file in your CVSROOT which corresponds to the file you were trying to look at, and send it to me. so I can evaluate it.
@dorothyr Zombie thread activate! (Sorry about that.)
Back in 2020, we sort of "moved full in to Git" and so my support requests (from users) for CVS stopped. However, recently, some old CVS code was revived and people asked about tkcvs. 2020 was a few OSs ago, so I installed tkrev anew (9.5) and decided to try again.
And so I tried again with the same file as above. Good news, doesn't crash. Bad news, doesn't work. I get a popup saying:
TkRev Warning!
Log empty. Check error status of cvs log command
Well, the file behaves for me, going back to tkcvs_8.0.4 which is as far back as my laptop will go.
Have you tried watching what happens in the trace window when you try to diagram the log? Turn on tracing, and just the Commands and Stderr levels might be enough to say what's going on. In this case, I accidentally got the error by re-generating my test repository and forgetting to hand-add the cvsxdiff file.
I think you did it right. Was stderr (E) logging on? But anyway cvs gave you the log. I wonder if there was an error status somewhere for some other reason.
Is this the only file you have a problem with?
Have you tried checking out a fresh copy of the module and starting over?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
@dorothyr Yeah, I had all the boxes checked in the window. And I just tried a fresh checkout with a different file and I get the same issue (see attached screenshot).
So, as the title says this is weird. I can use tkcvs in a Git repo just fine. But when I try to use it in a CVS repo:
I'm not sure how to decode that... Any help?
Note that this is with 9.3.2
And more info. If I just do
tkcvs
I get the list of files. But once I click the "tree" button, kaboom!I'll look at it. Thanks for posting the trace. I don't know what caused the branch browser to fail, but it's obvious that it didn't exit correctly.
That was interesting. The source code has been on the tigris.org collabnet ever since I made the Subversion module. Tigris closed down, and it took me all this time to get a dump so I could bring it back to Sourceforge.
Anyway, I fixed the proximate cause of the crash, which was this, but there was before that some reason why your revision-log file couldn't be processed. If it isn't proprietary, you could find the ,v file in your CVSROOT which corresponds to the file you were trying to look at, and send it to me. so I can evaluate it.
--- cvs.tcl (revision 960)
+++ cvs.tcl (working copy)
@@ -2700,7 +2700,7 @@
return
}
@dorothyr Zombie thread activate! (Sorry about that.)
Back in 2020, we sort of "moved full in to Git" and so my support requests (from users) for CVS stopped. However, recently, some old CVS code was revived and people asked about
tkcvs
. 2020 was a few OSs ago, so I installedtkrev
anew (9.5) and decided to try again.And so I tried again with the same file as above. Good news, doesn't crash. Bad news, doesn't work. I get a popup saying:
Well, okay:
I mean... no error.
I'm going to go ask someone who (I think) still has access to the CVS repo to see if he can pull out the ,v file.
Zombies! Somehow I'm still here though :-) It would be great to see the ,v file.
@dorothyr Okay. I'll work with the person with access to try and get it.
Note: I did try the usual "let's see if ChatGPT" can help and I'm not sure our robot friend knows TCL that well. :)
@dorothyr Please find attached the
cvsxdiff,v
file. Hopefullly it helps you figure out my oddity.Well, the file behaves for me, going back to tkcvs_8.0.4 which is as far back as my laptop will go.
Have you tried watching what happens in the trace window when you try to diagram the log? Turn on tracing, and just the Commands and Stderr levels might be enough to say what's going on. In this case, I accidentally got the error by re-generating my test repository and forgetting to hand-add the cvsxdiff file.
Here's the trace window
Hmm. How do you enter trace mode? I don't see a command line option.
Also: one thing I forgot. We access our CVS server via ssh (CVS_RSH=ssh) and we have to use a custom port (via .ssh/config). Could that cause issues?
Yes, access method can definitely cause problems. The trace should show that, hopefully.
In the top menubar, Options -> Tracing On/Off. In recent versions, you then
check the debug levels you want in the trace window's top bar.
On Fri, May 23, 2025 at 10:40 AM Matt Thompson theonematt@users.sourceforge.net wrote:
--
From my gmail account
Okay. Find attached a trace output. I think I did it right? But let me know if not.
I opened tkrev and pressed the "Branch" button for cvsxdiff. The warning popup came after:
I think you did it right. Was stderr (E) logging on? But anyway cvs gave you the log. I wonder if there was an error status somewhere for some other reason.
Is this the only file you have a problem with?
Have you tried checking out a fresh copy of the module and starting over?
@dorothyr Yeah, I had all the boxes checked in the window. And I just tried a fresh checkout with a different file and I get the same issue (see attached screenshot).