<arigato>
so a bare "git pull" only pulls the current branch, while a bare "git push" will try to push all branches and may succeed at some while failing at others
<arigato>
Good To Know
mathieu_ has joined #pypy
fotis has quit [Ping timeout: 240 seconds]
fotis has joined #pypy
<mathieu_>
@mattip FYI, I did look at the tools directory and found gcanalyze too but it does not do this. It parses logs generated by PYPYLOG=gc-collect:log.
<mathieu_>
@mattip FYI, I did spend some time tweaking my testcase and the extra memory allocations appear to be coming from the use of the \u2581 unicode char in re.sub('\s+', '\u2581', s). i.e., re.sub('\s+', ' ', s) does not use extra memory
<mathieu_>
so I looked in the sre implementation used in pypy and I see that it appears to have two codepaths, depending on the type of the arguments
<mathieu_>
I suspect I need to learn a bit about the pypy3 unicode support
<Dejan>
i need to write a tool that deletes all pypy wheels after i install new pypy :)
the_drow has joined #pypy
<mattip>
pip install --no-cache-dir ?
<Dejan>
i actually do want to freshen all pypy wheels
<Dejan>
will --no-cache-dir actually make pip overwrite in the cache?
fotis has quit [Ping timeout: 252 seconds]
fotis has joined #pypy
<mattip>
dunno
fotis has quit [Ping timeout: 252 seconds]
mathieu1 has joined #pypy
smarr has joined #pypy
<mathieu1>
[for the logs] FYI, I finally found out what was "wrong" with my testcase: I discovered that pypy stores unicode strings as utf-8 in memory, so the total memory usage depends on which exact characters appear and how often they appear and how many bytes each of them needs to be encoded as utf-8. Basically , re.sub('\s+', '\u2581') replaces a bunch of one-byte characters with a 3-byte character so,
<mathieu1>
re.sub increases my memory usage and the amount it increases is exactly equal to the number of times the regexp matches times (3-1)...
mathieu1 is now known as mathieu_
<smarr>
in RPython, is there an easy way to use `str.replace()` with a unicode character? It's a very basic thing: `name.replace(":", "⫶")` not essential, just cosmetic.
fotis has joined #pypy
<arigato>
smarr: yes, that should work, but if everything is unicode ('self' and the two arguments)
lritter has quit [Quit: Leaving]
<arigato>
(really unicodes, not utf-8 byte strings---if you're working with utf-8 byte strings then probably str.replace() cannot replace a byte with a multi-bytes character)
<ronan>
arigato, smarr: it should also work with str literals (the string will contain the utf8-encoded version), but in both cases you need to set the source file encoding to UTF-8
greedom has joined #pypy
<smarr>
thanks. and, right, I had marked the file as UTF-8. Simplest solution seems to be to chose another ASCII character and so I can avoid the issue.
fotis has quit [Ping timeout: 252 seconds]
greedom has quit [Remote host closed the connection]