May 28

Don’t do svn-to-git repository conversions with git-svn!

This is a public-service warning.

It has come to my attention that some help pages on the web are still recommending git-svn as a conversion tool for migrating Subversion repositories to git. DO NOT DO THIS. You may damage your history badly if you do.

Reminder: I am speaking as an expert, having done numerous large and messy repository conversions. I’ve probably done more Subversion-to-git lifts than anybody else, I’ve torture-tested all the major tools for this job, and I know their failure modes intimately. Rather more intimately than I want to…

Continue reading

May 18

Zeno tarpits

There’s a deeply annoying class of phenomena which, if you write code for any length of time, you will inevitably encounter. I have found it to be particularly prevalent in transformations to clean up or canonicalize large, complex data sets; repository export tools hit variants of it all the time, and so does my doclifter program for lifting [nt]roff markup to XML-DocBook.

It goes like this. You write code that handles a large fraction (say, 80%) of the problem space in a week. Then you notice that it’s barfing on the 20% remaining edge cases. These will be ugly to handle and greatly increase the complexity of your program, but it can be done, and you do it.

Once again, you have solved 80% of the remaining cases, and it took about a week – because your code is more complex than it used to be; testing it and making sure you don’t have regressions is about twice as difficult. But it can be done, at the cost of doubling your code complexity again, and you do it. Congratulations! You now handle 80% of the remaining cases. Then you notice that it’s barfing on 20% of remaining tricky edge cases….

…lather, rinse, repeat. If the problem space is seriously gnarly you can find yourself in a seemingly neverending cycle in which you’re expending multiplicatively more effort on each greater effort for multiplicatively decreasing returns. This is especially likely if your test range is expanding to include weirder data sets – in my case, older and gnarlier repositories or newer and gnarlier manual pages.

I think this is a common enough hazard of programming to deserve a name.

Continue reading

Apr 16

A belated response to “A Generation Lost in the Bazaar “

Back in 2012, Poul-Henning-Kamp wrote a disgruntled article in ACM Queue, A Generation Lost in the Bazaar.
It did not occur to me to respond in public at the time, but someone else’s comment on a G+ thread about the article revived the thread. Rereading my reaction, I think it is still worth sharing for the fundamental point about scaling and chaos.

Continue reading

Apr 02

My Gitorious projects have moved.

Gitorious – which I preferred to GitHub for being totally open-source – is shutting down sometime in May. I had no fewer than 26 projects on there, including reposurgeon, cvs-fast-import, doclifter, and INTERCAL.

Now they’ve moved. This won’t affect most of my users, as the web pages and distribution tarballs are still in their accustomed locations at If you’re a committer on any of these Gitirious repos, of course, the move actually matters.

Temporarily the repositories are on; here’s the entire list. They may not stay there, but moving them to was 90% of the work of moving them anywhere else and now I can consider options at my leisure.

Mar 08

Why I won’t mourn Mozilla

An incredibly shrinking Firefox faces endangered species status, says Computerworld, and reports their user market share at 10% and dropping. It doesn’t look good for the Mozilla Foundation – especially not with so much of their funding coming from Google which of course has its own browser to push.

I wish I could feel sadder about this. I was there at the beginning, of course – the day Netscape open-sourced the code that would become Mozilla and later Firefox was the shot heard ’round the world of the open source revolution, and the event that threw The Cathedral and the Bazaar into the limelight. It should be a tragedy – personally, for me – that the project is circling the drain.

Instead, all I can think is “They brought the fate they deserved on themselves.” Because principles matter – and in 2014 the Mozilla Foundation abandoned and betrayed one of the core covenants of open source.

Continue reading

Feb 22

GPSD 3.12 has shipped – bulletproofed from below

I’ve been radio silent the last couple of weeks mainly because I’ve been concentrating furiously on getting a GPSD release out the door. This one is a little more noteworthy than usual because it may actually have fixed a well-hidden flaw or vulnerability of some significance.

Regular readers may recall from back in 2013 that I published a heads-up titled No, GPSD is not the battery-killer on your Android! addressing a power-drain bug reported from a handful of Android phones.

I believed at the time that the proximate cause of the bug was in the kernel serial device-drivers somewhere specific to particular hardware on those phones. I still believe that, because if it had been a purely GPSD problem the error would likely have been much more widespread and I’d have been flooded with complaints.

However, I’ve been concerned ever since that GPSD might not have been doing everything it could to armor itself against bugginess in the layers below it. And a couple of weeks ago I found a problem…

Continue reading

Dec 13

Progress towards the extinction of CVS

The Great Beast, designed for converting large CVS repos, is now in full production. It hasn’t killed off any specimens in the wild yet (and I’ll explain why in a bit), but it’s doing spectacularly well on our test repositories.

As a representative large example, the entire Emacs CVS history, 1985-2009, 113309 CVS commits, lifts clean in 37 seconds at a sustained rate of 3K CVS commits a second. Yes, three thousand.

The biggest beast known to us, the NetBSD src repository, converts in 22 minutes. To give some idea of what a speedup this is, the first time I ran a lift on it – on one of Wendell’s Xeon machines – it took a bit under six hours. That’s about a factor of seventeen, there.

Judging by performance on the other project devs’ machines the Beast is good for a 2x to 3x speedup over a conventionally-balanced PC design (that is, one with worse RAM latency, narrower caches, more cores but somewhat lower single-thread speed). That’s a big enough advantage to validate the design and be practically significant on large repositories.

Continue reading

Nov 16

SRC 0.9: Ready for the less adventurous now

I just shipped SRC 0.9, and you no longer need to be adventurous to try it. It has a regression-test suite and real users.

Remarkably, SRC has had real users since 0.3, two days after it was born. Even more remarkably, the count of crash reports and botched operations from those users is zero. Zero. This is what you can gain from keeping code simple – I have has a couple of bug reports but they were both about filename quoting in the fast-export code, which is not a central feature.

Next, I’ll make a couple of what I think are important points about writing for zero defects. Then I’ll talk about a subtle issue or two in the design, and our one known behavioral glitch.

Continue reading

Nov 07

I wrote a version-control system today

I wrote a version-control system today. Yes, an entire VCS. Took me 14 hours.

Yeah, you’re looking at me like I’m crazy. “Why,” you ask, quite reasonably, “would you want to do a thing like that? We’re not short of powerful VCSes these days.

That is true. But I got to thinking, early this morning, about the fact that I haven’t been able to settle on just one VCS. I use git for most things, but there’s a use case git doesn’t cover. I have some document directories in which I have piles of things like HOWTOs which have separate histories from each other. Changes in them are not correlated, and I want to be able to move them around because I sometimes do that to reorganize them.

What have I been using for this? Why, RCS. The ancient Revision Control System, second oldest VCS in existence and clinging tenaciously to this particular niche. It does single-file change histories pretty well, but its UI is horrible. Worse than git’s, which is a pretty damning comparison.

Then I got to thinking. If I were going to design a VCS to do this particular single-file, single-user job, what would it look like? Hm. Sequential integer revision numbers, like Subversion and Mercurial used locally. Lockless operation. Modern CLI design. Built-in command help. Interchange with other VCSes via git import streams. This sounds like it could be nice

Then, the idea that made it inevitable. “I bet.” I thought, “I could write this thing as a Python wrapper around RCS tools. Use them for delta storage but hide all the ugly parts.”

Thus, SRC. Simple Revision Control, v0.1.

Continue reading

Nov 05

Chipping away at CVS

I’ve just shipped a new version of cvs-fast-export, 1.26. It speeds the tool up more, more, more – cranking through 25 years and 113300 commits of Emacs CVS history, for example in 2:48. That’s 672 commits a second, for those of you in the cheap seats.

But the real news this time is a Python wrapper called ‘cvsconvert’ that takes a CVS repository, runs a conversion to Git using cvs-fast-export, and then – using CVS for checkouts – examines the CVS and git repositories side by side looking for translation glitches. It checks every branch tip and every tag.

Running this on several of my test repos I’ve discovered some interesting things. One such discovery is of a bug in CVS. (Yeah, I know, what a shock…)

Continue reading

Oct 24

Moving the NetBSD repository

Some people on the NetBSD tech-repository list have wondered why I’ve been working on a full NetBSD repository conversion without a formal request from NetBSD’s maintainers that I do so.

It’s a fair question. An answer to it involves both historical contingency and some general issues about moving and mirroring large repositories. Because of the accident that a lot of people have recently dropped money on me in part to support an attack on this problem, I’m going to explain both in public.

Continue reading

Oct 20

Building the perfect beast

I’ve attempted to summarize the discussion of build options for the repository-surgery machine. You should see a link at the top of the page: if not, it’s here

I invite all the commenters who have shown an interest to critique these build proposals. Naturally, I’d like to make sure we have a solid parts list with no spec conflicts before we start spending money and time to build this thing.

Continue reading

Oct 18

Black magic and the Great Beast

Something of significance to the design discussion for the Great Beast occurred today.

I have finally – finally! – achieved significant insight into the core merge code, the “black magic” section of cvs-fast-export. If you look in merge.c in the repo head version you’ll see a bunch of detailed comments that weren’t there before. I feel rather as Speke and Burton must have when after weeks of hacking their way through the torrid jungles of darkest Africa they finally glimpsed the source of the Nile…

Continue reading

Oct 16

A low-performance mystery: Sometimes you gotta simplify

This series of posts is increasingly misnamed, as there is not much mystery left about cvs-fast-export’s performance issues and it is now blazingly, screamingly, bat-out-of-hell fast. As in both threaded and unthreaded version convert the entire history of groff (15593 CVS deltas in 1549 files in 13 seconds flat. That would be about 10K CVS commits per minute, sustained; in practice the throughput will probably fall off a bit on very large repositories.

I achieved the latest doubling in speed by not succumbing to the temptation to overengineer – a trap that lays in wait for all clever hackers. Case study follows.

Continue reading

Oct 14

A low-performance mystery: the adventure continues

The mystery I described two posts back has actually been mostly solved (I think) but I’m having a great deal of fun trying to make cvs-fast-export run even faster, and my regulars are not only kibitzing with glee but have even thrown money at me so I can upgrade my PC and run tests on a machine that doesn’t resemble (as one of them put it) a slack-jawed yokel at a hot-dog-eating contest.

Hey, a 2.66Ghz Intel Core 2 Duo with 4GB was hot shit when I bought it, and because I avoid bloatware (my window manager is i3) it has been sufficient unto my needs up to now. I’m a cheap bastard when it comes to hardware; tend to hold onto it until I actually need to upgrade. This is me loftily ignoring the snarking from the peanut gallery, except from the people who actually donated money to the Help Stamp Out CVS In Your Lifetime hardware fund.

(For the rest of you, the PayPal and Gratipay buttons should be clearly visible to your immediate right. Just sayin’…)

Ahem. Where was I? Yes. The major mystery – the unexplained slowdown in stage 3 of the threaded version – appears to have been solved. It appears this was due to a glibc feature, which is that if you link with threads support it tries to detect use of threads and use thread locks in stdio to make it safe. Which slows it down.

Continue reading

Oct 12

A low-performance mystery, part deux

Well, the good news is, I get to feel wizardly this morning. Following sensible advice from a couple of my regulars, I rebuilt my dispatcher to use threads allocated at start time and looping until the list of masters is exhausted.

78 LOC. Fewer mutexes. And it worked correctly first time I ran it. W00t – looks like I’ve got the hang non-hang of this threads thing.

The bad news is, threaded performance is still atrocious in exactly the same way. Looks like thread-spawn overhead wasn’t a significant contributor.

In truth, I was expecting this result. I think my regulars were right to attribute this problem to cache- and locality-busting on every level from processor L1 down to the disks. I believe I’m starting to get a feel for this problem from watching the performance variations over many runs.

I’ll profile, but I’m sure I’m going to see cache misses go way up in the threaded version, and if I can find a way to meter the degree of disk thrashing I won’t be even a bit surprised to see that either.

The bottom line here seems to be that if I want better threaded performance out of this puppy I’m going to have to at least reduce its working set a lot. Trouble is, I’m highly doubtful – given what it has to do during delta assembly – that this is actually possible. The CVS snapshots and deltas it has to snarf into memory to do the job are intrinsically both large and of unpredictably variable size.

Maybe I’ll have an inspiration, but…Keith Packard, who originally wrote that code, is a damn fine systems hacker who is very aware of performance issues; if he couldn’t write it with a low footprint in the first place, I don’t judge my odds of second-guessing him successfully are very good.

Ah well. It’s been a learning experience. At least now I can say of multi-threaded application designs “Run! Flee! Save yourselves!” from a position of having demonstrated a bit of wizardry at them myself.

UPDATE: One of my regulars found a minor bug in the mutex handling that cost some performance. Alas, fixing this didn’t have any impact above the noise level of my profiling. Also, I managed to unify the threaded and non-threaded dispatchers; the LOC specific to threading is now down to about 30.

Oct 11

A low-performance mystery

OK, I’ll admit it. I’m stumped by a software-engineering problem.

This is not a thing that happens often, but I’m in waters relatively unknown to me. I’ve been assiduously avoiding multi-threaded programming for a long time, because solving deadlock, starvation, and insidious data-corruption-by-concurrency problems isn’t really my idea of fun. Other than one minor brush with it handling PPS signals in GPSD I’ve managed before this to avoid any thread-entanglement at all.

But I’m still trying to make cvs-fast-export run faster. About a week ago an Aussie hacker named David Leonard landed a brilliant patch series in my mailbox. Familiar story: has a huge, gnarly CVS repo that needs converting, got tired of watching it grind for days, went in to speed it up, found a way. In fact he applied a technique I’d never heard of (Bloom filtering) to flatten the worst hot spot in the code, an O(n**3) pass used to compute parent/child links in the export code. But it still needed to be faster.

After some discussion we decided to tackle parallelizing the code in the first stage of analysis. This works – separately – on each of the input CVS masters, digesting them into in-core revision lists and generating whole-file snapshots for each CVS delta; later these will become the blobs in the fast-export stream. Then there’s a second stage that merges these per-file revision lists, and a third stage that exports the merged result.

Here’s more detail, because you’ll need it to understand the rest. Each CVS master consists of a sequence of deltas (sequences of add-line and delete-line operations) summing up to a sequence of whole-file states (snapshots – eventually these will become blobs in the translated fast-import-stream). Each delta has an author, a revision date, and a revision number (like 1.3 or Implicitly they form a tree. At the top of the file is a tag table mapping names to revision numbers, and some other relatively unimportant metadata.

The goal of stage 1 is to digest each CVS master into an in-core tree of metadata and a sequence of whole-file snapshots, with unique IDs in the tree indexing the snapshots. The entire collection of masters is made into a linked list of these trees; this is passed to stage 2, where black magic that nobody understands happens.

This first stage seems like a good target for parallelization because the analysis of each master consists of lumps of I/O separated by irregular stretches of compute-intensive data-shuffling in core. In theory, if the program were properly parallelized, it would seldom actually block on an I/O operation; instead while any one thread was waiting on I/O, the data shuffling for other masters would continue. The program would get faster – possibly much faster, depending on the time distribution of I/O demand.

Well, that’s the theory, anyway. Here’s what actually happened…

Continue reading