From: SourceForge.net <no...@so...> - 2008-06-16 19:14:50
|
Bugs item #1995305, was opened at 2008-06-16 18:34 Message generated for change (Comment added) made by jordiflor You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=833326&aid=1995305&group_id=164845 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: None Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Jordi Floriach (jordiflor) Assigned to: Nobody/Anonymous (nobody) Summary: Wrong lines of Code calculation Initial Comment: subversion-1.4.4-30 statsvn-0.4.0 Hello, stats made on different revisions show different lines of code on past commits. For example, I checked out rev 42 and 80 and generate stats for each one (using full log). For a given file, I get these dumps: --- Revision 42 File 11/ src/foo.c LOC:687 Revision:27 Delta:878 Lines:878 Ini::Y BegLog:N Dead:N SumDelta:878 null Revision:31 Delta:-84 Lines:794 Ini::N BegLog:N Dead:N SumDelta:794 null Revision:34 Delta:0 Lines:794 Ini::N BegLog:N Dead:N SumDelta:794 null Revision:36 Delta:-2 Lines:792 Ini::N BegLog:N Dead:N SumDelta:792 null Revision:37 Delta:-105 Lines:687 Ini::N BegLog:N Dead:N SumDelta:687 null Revision:39 Delta:0 Lines:687 Ini::N BegLog:N Dead:N SumDelta:687 null Revision:41 Delta:0 Lines:687 Ini::N BegLog:N Dead:N SumDelta:687 null --- Revision 80 File 11/ src/foo.c LOC:801 Revision:27 Delta:1143 Lines:1143 Ini::Y BegLog:N Dead:N SumDelta:1143 null Revision:31 Delta:-84 Lines:1059 Ini::N BegLog:N Dead:N SumDelta:1059 null Revision:34 Delta:0 Lines:1059 Ini::N BegLog:N Dead:N SumDelta:1059 null Revision:36 Delta:-2 Lines:1057 Ini::N BegLog:N Dead:N SumDelta:1057 null Revision:37 Delta:-105 Lines:952 Ini::N BegLog:N Dead:N SumDelta:952 null Revision:39 Delta:0 Lines:952 Ini::N BegLog:N Dead:N SumDelta:952 null Revision:41 Delta:0 Lines:952 Ini::N BegLog:N Dead:N SumDelta:952 null Revision:43 Delta:1 Lines:953 Ini::N BegLog:N Dead:N SumDelta:953 null Revision:44 Delta:-64 Lines:889 Ini::N BegLog:N Dead:N SumDelta:889 null Revision:45 Delta:-86 Lines:803 Ini::N BegLog:N Dead:N SumDelta:803 null Revision:46 Delta:45 Lines:848 Ini::N BegLog:N Dead:N SumDelta:848 null Revision:47 Delta:14 Lines:862 Ini::N BegLog:N Dead:N SumDelta:862 null Revision:50 Delta:-9 Lines:853 Ini::N BegLog:N Dead:N SumDelta:853 null Revision:55 Delta:-69 Lines:784 Ini::N BegLog:N Dead:N SumDelta:784 null Revision:56 Delta:-45 Lines:739 Ini::N BegLog:N Dead:N SumDelta:739 null Revision:59 Delta:-13 Lines:726 Ini::N BegLog:N Dead:N SumDelta:726 null Revision:60 Delta:34 Lines:760 Ini::N BegLog:N Dead:N SumDelta:760 null Revision:61 Delta:0 Lines:760 Ini::N BegLog:N Dead:N SumDelta:760 null Revision:63 Delta:31 Lines:791 Ini::N BegLog:N Dead:N SumDelta:791 null Revision:65 Delta:54 Lines:845 Ini::N BegLog:N Dead:N SumDelta:845 null Revision:66 Delta:35 Lines:880 Ini::N BegLog:N Dead:N SumDelta:880 null Revision:68 Delta:-13 Lines:867 Ini::N BegLog:N Dead:N SumDelta:867 null Revision:71 Delta:-29 Lines:838 Ini::N BegLog:N Dead:N SumDelta:838 null Revision:75 Delta:-37 Lines:801 Ini::N BegLog:N Dead:N SumDelta:801 null The fact is that in rev 27 (this file's initial) file has 677 lines. Every stat with a greater revision increases the original size of the file (and this applies to all files). I cannot figure out how this numbers are calculated. The fact is that LOC graphics each time has a bigger scale on past events. Final number of lines is correct, but initial and deltas aren't. Regards, Jordi Floriach ---------------------------------------------------------------------- >Comment By: Jordi Floriach (jordiflor) Date: 2008-06-16 21:14 Message: Logged In: YES user_id=2069575 Originator: YES Hi, I apologize for making you loss your time. It looked like so awful bug that it wasn't possible that nobody had submitted it before. Problem was originated in a "missing revision" due to some irregularity in cvs2svn process. I've recreated SVN repository again and all stuff looks pretty OK. Thanks, Jordi ---------------------------------------------------------------------- Comment By: Jason Kealey (jkealey) Date: 2008-06-16 19:56 Message: Logged In: YES user_id=1498924 Originator: NO You may have found a real bug here. I wonder if we even use those past values anywhere in the reports ... (Haven't tried -dump on my project yet.). I think we only work with the diffs and final counts, never the actual past line counts. (to be investigated if this is a bug in the -dump or a bug in the data model) By the way, expect a few data inconsistencies because we don't support branch svn move or copy (they are seen as deletion+addition or addition). The current results aren't perfect, but they give a good idea of what is going on. ---------------------------------------------------------------------- Comment By: Jordi Floriach (jordiflor) Date: 2008-06-16 19:51 Message: Logged In: YES user_id=2069575 Originator: YES Thanks for fast answer. Now I figured out... deltas are well calculated, but applied backwards with opposite sign and on the n-1 affected revision. These are linecounts (wc -l) for revisions with changes, resulting deltas match with dump, but the sign: 878 r63/src/foo.c 847 r65/src/foo.c +31 793 r66/src/foo.c -54 758 r68/src/foo.c -35 771 r71/src/foo.c +13 800 r75/src/foo.c +29 Now the question is: is this problem reproducible or only affects to me? Regards, Jordi ---------------------------------------------------------------------- Comment By: Jason Kealey (jkealey) Date: 2008-06-16 18:48 Message: Logged In: YES user_id=1498924 Originator: NO Interesting. What does the svn diff of foo.c between revisions 66 and 68 give you, for example. (FYI, we start with the current line count in the working copy, and work our way back using the log and svn diff.) ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=833326&aid=1995305&group_id=164845 |