index.html

   1 <html>
   2         <head>
   3                 <title>Dscho's Git log</title>
   4                 <meta http-equiv="Content-Type"
   5                         content="text/html; charset=UTF-8"/>
   6         </head>
   7         <body style="width:800px;background-image:url(dscho.git?a=blob_plain;hb=832be85c785c80202f17b87db7f063ae57ec2cac;f=paper.jpg);background-repeat:repeat-y;background-attachment:scroll;padding:0px;">
   8                 <div style="width:610px;margin-left:120px;margin-top:50px;align:left;vertical-align:top;">
   9                         <h1>Dscho's Git log</h1>
  10                         <div style="position:absolute;top:50px;left:810px;width=400px">
  11                         <table width=400px bgcolor=#e0e0e0 border=1>
  12                         <tr><th>Table of contents:</th></tr>
  13                         <tr><td>
  14                         <p><ul>
  15                         <li><a href=#1236479389>08 Mar 2009 New Git for Windows version</a>
  16                         <li><a href=#1235092615>20 Feb 2009 Code reviews</a>
  17                         <li><a href=#1234409395>12 Feb 2009 Interactive <i>rebase</i> just learnt a new command: <i>topic</i></a>
  18                         <li><a href=#1234320806>11 Feb 2009 Thunderbird, oh Thunderbird, you always make my small brain hurt</a>
  19                         <li><a href=#1234141489>09 Feb 2009 <i>format-patch --thread</i> and Alpine</a>
  20                         <li><a href=#1234140696>09 Feb 2009 <i>rebase</i> updates</a>
  21                         <li><a href=#1234040744>07 Feb 2009 The infamous <i>mark</i> command in the <i>rebase</i> command</a>
  22                         <li><a href=#1233707628>04 Feb 2009 New valgrind series</a>
  23                         <li><a href=#1233706294>04 Feb 2009 Problems with split-topic-branches.sh</a>
  24                         <li><a href=#1233277286>30 Jan 2009 More valgrind fun</a>
  25                         </ul></p>
  26                         <a href=dscho.git?a=blob_plain;hb=e0559e8e91d1c01a337b56a3ff00234a33106924;f=index.html>Older posts</a>
  27                         </td></tr></table>
  28                         <br>
  29                         <div style="text-align:right;">
  30                         <a href="dscho.git?a=blob_plain;hb=blog;f=blog.rss"
  31                            title="Subscribe to my RSS feed"
  32                            class="rss" rel="nofollow"
  33                            style="background-color:orange;text-decoration:none;color:white;font-family:sans-serif;">RSS</a>
  34                         </div>
  35                         <br>
  36                         <table width=400px bgcolor=#e0e0e0 border=1>
  37                         <tr><th>About this blog:</th></tr>
  38                         <tr><td>
  39                         <p>It is an active <a href=http://repo.or.cz/w/git/dscho.git?a=blob_plain;f=index.html;hb=5f002cab57a837125a8f901bcd1f3c1477bc3119>abuse</a> of <a href=http://repo.or.cz/>repo.or.cz</a>,
  40                         letting gitweb unpack the objects in the current tip of the branch <i>blog</i>,
  41                         including the images and the RSS feed.
  42                         </p><p>
  43                         Publishing means running a script that collects the posts, turns them into
  44                         HTML, makes sure all the images are checked in, and pushes the result.
  45                         </p><p>
  46                         This blog also serves to grace the world with Dscho's random thoughts on and
  47                         around Git.
  48                         </p>
  49                         </td></tr></table>
  50                         <br>
  51                         <table width=400px bgcolor=#e0e0e0 border=1>
  52                         <tr><th>Links:</th></tr>
  53                         <tr><td>
  54                         <ul>
  55                         <li> <a href=http://git-scm.com/>Git's homepage</a>
  56                         <li> <a href=http://gitster.livejournal.com/>Junio's blog</a>
  57                         <li> <a href=http://www.spearce.org/>Shawn's blog</a> seems to be sitting
  58                               idle ever since he started working for Google...
  59                         <li> <a href=http://torvalds-family.blogspot.com/>Linus' blog</a> does not
  60                               talk much about Git...
  61                         <li> Scott Chacon's <a href=http://whygitisbetterthanx.com/>Why Git is better
  62                              than X</a> site
  63                         <li> <a href=http://vilain.net/>The blog of mugwump</a>
  64                         <li> <a href=http://blogs.gnome.org/newren/>Elijah Newren</a> chose the
  65                               same path as Cogito, offering an alternative porcelain (an approach
  66                               that is doomed in my opinion)
  67                         <li> <a href=http://msysgit.googlecode.com/>The msysGit project</a>, a (mostly)
  68                               failed experiment to lure the many Windows developers out there to
  69                               contribute to Open Source for a change.
  70                         </ul>
  71                         </td></tr></table>
  72                         <br>
  73                         <table width=400px bgcolor=#e0e0e0 border=1>
  74                         <tr><th>Google Ads:</th></tr>
  75                         <tr><td>
  76                         <script type="text/javascript"><!--
  77                         google_ad_client = "pub-5106407705643819";
  78                         /* 300x250, created 1/22/09 */
  79                         google_ad_slot = "6468207338";
  80                         google_ad_width = 300;
  81                         google_ad_height = 250;
  82                         //-->
  83                         </script>
  84                         <script type="text/javascript"
  85                         src="http://pagead2.googlesyndication.com/pagead/show_ads.js">
  86                         </script>
  87                         </td></tr></table>
  88                         </div>
  89                         <h6>Sunday, 8th of March, Anno Domini MMIX, at the hour of the Tiger</h6>
  90                         <a name=1236479389>
  91                         <h2>New Git for Windows version</h2>
  92
  93                         <p>
  94                         </p><p>
  95                         Phew.  That was quite a day, almost exclusively spent on finishing that
  96                         installer.  The worst part: updating GCC seemed not to be such a good idea
  97                         after all...
  98                         </p><p>
  99                         For Windows, we need to use the printf format <i>%I64u</i> (which is
 100                         non-standard, in the common way of Microsoft) if you want to print 64-bit
 101                         wide unsigned numbers.  The rest of the world accepts the standard <i>%llu</i>.
 102                         </p><p>
 103                         After upgrading to the new GCC, a lot of warnings appeared, complaining
 104                         about <i>%I64u</i>.  The warnings went away when I replaced the format with
 105                         <i>%llu</i>.
 106                         </p><p>
 107                         Being the naive I am, I mistook that for a sign that we could finally go
 108                         more standards-compliant.
 109                         </p><p>
 110                         However, it only means that we have to live with the warnings for now, as
 111                         the C runtime provided on Windows still strongly disagrees with standards
 112                         (and it has to continue to do so, lest it break existing programs).
 113                         </p><p>
 114                         Sigh.
 115                         </p><p>
 116                         At least I have the feeling that I caught the most important bugs before
 117                         releasing.
 118                         </p>
 119                         <h6>Friday, 20th of February, Anno Domini MMIX, at the hour of the Buffalo</h6>
 120                         <a name=1235092615>
 121                         <h2>Code reviews</h2>
 122
 123                         <p>
 124                         </p><p>
 125                         It has been said that reviewing patches is a most thankless job.  As I really
 126                         like the elegance of Git's source code, and care a lot about it, I did not
 127                         think that it was thankless, just a little bit tedious (especially when the
 128                         patch authors mistake criticism for personal attacks).
 129                         </p><p>
 130                         Usually, I am pretty good at ignoring insults as responses to my comments;
 131                         after all, I have a lot more enjoyable things to do than to spend time talking
 132                         to a guy who shows how wise he is when he thinks that I criticize him
 133                         <u>personally</u> when I just try to enhance his work, by offering a little bit of
 134                         my knowledge.
 135                         </p><p>
 136                         However, in the last days, three people really seemed to want to insult me,
 137                         to make me go away, to stop the fun I have with Git.
 138                         </p><p>
 139                         And they almost succeeded.
 140                         </p><p>
 141                         So I guess it is time to reassess my priorities, and maybe stop reviewing
 142                         Git patches altogether.
 143                         </p>
 144                         <h6>Thursday, 12th of February, Anno Domini MMIX, at the hour of the Tiger</h6>
 145                         <a name=1234409395>
 146                         <h2>Interactive <i>rebase</i> just learnt a new command: <i>topic</i></h2>
 147
 148                         <p>
 149                         </p><p>
 150                         Today I am pretty pleased with myself.  Two projects at my day job got a real
 151                         boost, and I implemented a shortcut that avoids the ugly 'bookmark' statement
 152                         in rebase scripts most of the time.
 153                         </p><p>
 154                         A typical rebase script, generated by <i>git rebase -i -p $COMMIT</i> will look
 155                         something like this:
 156                         </p><p>
 157                         <table
 158                                                         border=1 bgcolor=black>
 159                                                 <tr><td bgcolor=lightblue colspan=3>
 160                                 <pre>                                                                                </pre>
 161                         </td></tr>
 162                         <tr><td>
 163                                 <table cellspacing=5 border=0
 164                                          style="color:white;">
 165                                 <tr><td>
 166                                         <pre>
 167 pick 1234567 My first commit
 168 topic begin super-cool-feature
 169 pick 2345678 The super cool feature
 170 pick 3456789 Documentation for the super cool feature
 171 topic end super-cool-feature
 172 </pre>
 173                                                         </td></tr>
 174                                                         </table>
 175                                                 </td></tr>
 176                                                 </table>
 177                         </p><p>
 178                         The result will be a merge commit at the HEAD whose first parent is
 179                         "My first commit", whose second parent is "Documentation for the super
 180                         cool feature" and whose commit message is "Merge branch 'super-cool-feature'".
 181                         </p><p>
 182                         Side note: internally, <i>topic begin $NAME [at $COMMIT]</i> will be handled as if
 183                         you wrote <i>bookmark merge-parent-of-$NAME; goto $COMMIT</i>, and
 184                         <i>topic end $NAME [$MESSAGE]</i> will be handled as if you wrote
 185                         <i>bookmark $NAME; goto merge-parent-of-$NAME; merge parents $NAME [original $MARK Merge branch '$NAME']</i>.
 186                         </p><p>
 187                         Of course, being more concise, the 'topic' statement is not only nicer to the
 188                         eye, but also less error-prone.
 189                         </p><p>
 190                         And hopefully many people will agree with me that this rebase script is pretty
 191                         intuitive.
 192                         </p>
 193                         <h6>Wednesday, 11th of February, Anno Domini MMIX, at the hour of the Tiger</h6>
 194                         <a name=1234320806>
 195                         <h2>Thunderbird, oh Thunderbird, you always make my small brain hurt</h2>
 196
 197                         <p>
 198                         </p><p>
 199                         There was a lengthy discussion on the Git mailing list about using Thunderbird,
 200                         a not quite unpopular mailing program, to send inline patches.
 201                         </p><p>
 202                         It is really kind of sad that the Thunderbird developers do not see how
 203                         stubbornly they offend quite a number of people and scare them away from their
 204                         program.  After all, you should try to be liberal in what you accept and strict
 205                         in what you emit.  No, that does not mean that you should force others to
 206                         switch their mailers because you strictly adher to your philosophy in what you
 207                         emit, ignoring the rest of the world.
 208                         </p><p>
 209                         In any case, I am not affected (as long as I do not get mails from a poor soul
 210                         stuck with Thunderbird).
 211                         </p><p>
 212                         But I was a bit mean to that Thunderbird guy I dragged into the discussion, and
 213                         he seems really offended.
 214                         </p><p>
 215                         So I thought I'd give him a real reason to feel offended: I'll just do his work:
 216                         </p><p>
 217                         http://<a href=http://repo.or.cz>repo.or.cz</a>/w/UnFlowedThunderbird.git
 218                         </p><p>
 219                         It took my free time of two days, being not a Thunderbird developer myself.
 220                         Hopefully it works, and hopefully some people will feel really ashamed now.
 221                         </p>
 222                         <h6>Monday, 9th of February, Anno Domini MMIX, at the hour of the Buffalo</h6>
 223                         <a name=1234141489>
 224                         <h2><i>format-patch --thread</i> and Alpine</h2>
 225
 226                         <p>
 227                         </p><p>
 228                         I started recently to pipe the output of
 229                         <i>git format-patch --cover-letter --stdout ...</i> directly into the
 230                         <i>postponed-msgs</i> folder Alpine uses, instead of pasting files into the
 231                         mailer.
 232                         </p><p>
 233                         The idea is to pretend that I continue a postponed mail, but in reality I
 234                         never wrote it, <i>format-patch</i> did.
 235                         </p><p>
 236                         However, I had problems with the <i>--thread</i> option that is implied by
 237                         <i>--cover-letter</i>.  Alpine always generated new message IDs without adjusting
 238                         the <i>In-reply-to:</i> and <i>References:</i> headers of the other mails.
 239                         </p><p>
 240                         Now I found out that the reason is that the <i>Fcc:</i> headers were missing in
 241                         the mails, and Alpine generated them, making up new message IDs in the process.
 242                         </p><p>
 243                         Therefore I have an alias now which sets not only the <i>Fcc:</i> header, but also
 244                         the <i>To:</i> headers by rewriting the stream using <i>sed</i>.  This is slightly
 245                         ugly, but so is the handling of headers in <i>format-patch</i>: if you thought
 246                         you could specify arbitrary headers using the command line, you are mistaken:
 247                         you can do that only by editing the config.
 248                         </p><p>
 249                         While at it, I also noticed a bug whereby <i>--thread --in-reply-to=...</i> simply
 250                         forgets the <i>--thread</i>.  Maybe this week I will find time to address this bug.
 251                         </p>
 252                         <h6>Monday, 9th of February, Anno Domini MMIX, at the hour of the Buffalo</h6>
 253                         <a name=1234140696>
 254                         <h2><i>rebase</i> updates</h2>
 255
 256                         <p>
 257                         </p><p>
 258                         Phew.  The last few days, I was mainly chasing bugs I introduced due to being
 259                         too tired to work on the merge-preserving, interactive <i>rebase</i>.
 260                         </p><p>
 261                         But finally I have something I can start working with.  After my failed
 262                         experiment to use git-blame to split topic branches, I will sort the commits
 263                         in my <i>my-next</i> branch into topic branches manually.
 264                         </p><p>
 265                         Then I will add an option to <i>rebase -i -p</i> to rewrite refs which point to
 266                         rewritten commits, so that I can have branches <i>rebase-i-p</i>, <i>add-e</i>, etc
 267                         and all of them are automatically updated when I <i>rebase -i -p</i> the <i>my-next</i>
 268                         branch.
 269                         </p><p>
 270                         In the process, not only have I learnt the value of the <i>bookmark</i> command,
 271                         but made quite a few-much needed cleanups (which make
 272                         <i>git-rebase--interactive.sh</i> longer, but much more understandable).
 273                         </p><p>
 274                         Hopefully Stephan will pick the changes in the "rebase protocol" up, and then
 275                         we can have a sequencer with which I can start to make a graphical interactive
 276                         rebase using git-gui.  Or gitk.
 277                         </p><p>
 278                         Maybe.
 279                         </p>
 280                         <h6>Saturday, 7th of February, Anno Domini MMIX, at the hour of the Pig</h6>
 281                         <a name=1234040744>
 282                         <h2>The infamous <i>mark</i> command in the <i>rebase</i> command</h2>
 283
 284                         <p>
 285                         </p><p>
 286                         I realized today how easy it is to lose commits with the "merge preserving"
 287                         mode of the interactive rebase.  In my case, it was when I tried to move a
 288                         bunch of commits from the tip of my branch into a topic branch.
 289                         </p><p>
 290                         But after moving the commits, I forgot to update the parent of the merge
 291                         commit.  Possibly a mark command could have helped.  The very same command
 292                         I called a nightmare for usability.
 293                         </p><p>
 294                         So I was wrong.  Big news.  &#x263a;
 295                         </p><p>
 296                         However, I think that the syntax "mark :1" is something best left for
 297                         machine consumption, not for human beings.
 298                         </p><p>
 299                         But I have an idea: we could use some garbled commit subject, or in case of
 300                         merge parents, the merge subject as some human readable title of the mark.
 301                         </p><p>
 302                         The rebase script would then look something like this:
 303                         </p><p>
 304                         <table
 305                                                         border=1 bgcolor=black>
 306                                                 <tr><td bgcolor=lightblue colspan=3>
 307                                 <pre>                                                                                </pre>
 308                         </td></tr>
 309                         <tr><td>
 310                                 <table cellspacing=5 border=0
 311                                          style="color:white;">
 312                                 <tr><td>
 313                                         <pre>
 314 pick abcdefg Some ultra cool commit
 315 bookmark ultra-cool
 316 goto upstream
 317 pick hijklmn Some other cool commit
 318 merge parent ultra-cool Merge 'ultra-cool' into master
 319 </pre>
 320                                                         </td></tr>
 321                                                         </table>
 322                                                 </td></tr>
 323                                                 </table>
 324                         </p><p>
 325                         The good news is: I added code that refuses to finish a rebase when there
 326                         are commits that were rewritten, but not part of the new HEAD's ancestry.
 327                         </p>
 328                         <h6>Wednesday, 4th of February, Anno Domini MMIX, at the hour of the Buffalo</h6>
 329                         <a name=1233707628>
 330                         <h2>New valgrind series</h2>
 331
 332                         <p>
 333                         </p><p>
 334                         I spent quite some time cleaning up that patch series, and feel pretty
 335                         exhausted.
 336                         </p><p>
 337                         Granted, the new <i>git rebase -i -p</i> does its job without complaint so far
 338                         (so much so that I think I'll release a version of my <i>rebase</i> series
 339                         soonish), but it <u>is</u> a hassle when you have patches that you have a hard
 340                         time to decide upon the order/commit boundaries.
 341                         </p><p>
 342                         For example, I could imagine that the patch making the location of the
 343                         templates independent of the location of the Git binaries should come
 344                         <u>before</u> my patch series, and the valgrind specific part should then
 345                         be squashed into the first valgrind commit.
 346                         </p><p>
 347                         Also, it uses two features of valgrind 3.4.0:
 348                         </p><p>
 349                         <ul>
 350                         <li><i>...</i> in the suppression file, and
 351                         <li><i>--track-origins=yes</i>
 352                         </ul>
 353                         </p><p>
 354                         The latter is actually the reason I am pretty willing to keep the
 355                         requirement of that valgrind version, as it is really, really useful.
 356                         </p><p>
 357                         I guess we will see what happens to it.
 358                         </p>
 359                         <h6>Wednesday, 4th of February, Anno Domini MMIX, at the hour of the Buffalo</h6>
 360                         <a name=1233706294>
 361                         <h2>Problems with split-topic-branches.sh</h2>
 362
 363                         <p>
 364                         </p><p>
 365                         So my little script that should help me to split my topic branches does
 366                         not work properly.
 367                         </p><p>
 368                         First some background: the idea was to let <i>git blame</i> do the hard work
 369                         to find overlapping changes, i.e. changes that would conflict when
 370                         changing the order (or skipping the first change, on which the next builds).
 371                         </p><p>
 372                         The first problem with that approach: when lines are <u>removed</u> by one
 373                         commit, and the next commit touches the same location, <i>git blame</i> does
 374                         not find that the first commit is required by the second.
 375                         </p><p>
 376                         Therefore I introduced a really slow reverse thing which tries to find
 377                         those commits whose removals survived until the parent of a particular
 378                         commit, but not further.
 379                         </p><p>
 380                         However, it does not work properly.  Basically, only context sizes that
 381                         span the whole files lead to conflict-free topic branches so far.
 382                         </p><p>
 383                         As a consequence, I think I'll add an option --sprout to the revision
 384                         walker which will fake octopus merges (or a series of two-parent merges)
 385                         whenever it finds a perl of non-merge commits that are theoretically
 386                         independent, i.e. whose patches apply cleanly.
 387                         </p>
 388                         <h6>Friday, 30th of January, Anno Domini MMIX, at the hour of the Buffalo</h6>
 389                         <a name=1233277286>
 390                         <h2>More valgrind fun</h2>
 391
 392                         <p>
 393                         </p><p>
 394                         So I spent quite a number of hours on that funny zlib/valgrind issue.  The
 395                         thing is, zlib people claim that even if their code accesses uninitialized
 396                         memory, it does not produce erroneous data (by cutting out the results of the
 397                         uninitialized data, which is cheaper than checking for the end of the buffer
 398                         in an unaligned manner), so zlib will always be special for valgrind.
 399                         </p><p>
 400                         However, the bug I was chasing is funny, and different from said issue.  zlib
 401                         deflates an input buffer to an output buffer that is exactly 58 bytes long.
 402                         But valgrind claims that the 52nd of those bytes is uninitialized, and <u>only</u>
 403                         that one.
 404                         </p><p>
 405                         But it is not.  It must be 0x2c, otherwise zlib refuses to inflate the
 406                         buffer.
 407                         </p><p>
 408                         Now, I went into a debugging frenzy, and finally found out that zlib just
 409                         passes fine (with the default suppressions because of the "cute" way it
 410                         uses uninitialized memory), <u>except</u> when it is compiled with UNALIGNED_OK
 411                         defined.
 412                         </p><p>
 413                         Which Ubuntu does, of course.  Ubuntu, the biggest forker of all.
 414                         </p><p>
 415                         The bad part is that it sounds like a bug in valgrind, and I <u>could</u> imagine
 416                         that it is an issue of an optimized memcpy() that copies int by int, and
 417                         that valgrind misses out on the fact that a part of that int is actually
 418                         <u>not</u> uninitialized.
 419                         </p><p>
 420                         But my debugging session's results disagree with that.
 421                         </p><p>
 422                         With the help of Julian Seward, the original author of valgrind, I instrumented
 423                         zlib's source code so that valgrind checks earlier if the byte is initialized
 424                         or not, to find out where the reason of the issue lies.
 425                         </p><p>
 426                         The sad part is that when I added the instrumentation to both the <u>end</u> of
 427                         the while() loop in compress_block() in zlib's trees.c, and just <u>after</u> the
 428                         while() loop (whose condition is a plain <i>variable < variable</i> comparison,
 429                         nothing fancy, certainly not changing any memory), only the <u>latter</u> catches
 430                         a valgrind error.
 431                         </p><p>
 432                         And that is truly strange.
 433                         </p>
 434                 </div>
 435         </body>
 436 </html>