[Cocci] prequel

[Cocci] prequel

[Julia Lawall]

Prequel is like Coccinelle, but for searching in Git commit histories.

The source code of Prequel v0.1 is available via the following web page:

http://prequel-pql.gforge.inria.fr/

Feedback is welcome.

julia

[Cocci] Querying Git commits with Prequel

[SF Markus Elfring]

> Prequel is like Coccinelle, but for searching in Git commit histories.

I am curious on how this software approach will evolve further.


> http://prequel-pql.gforge.inria.fr/
> 
> Feedback is welcome.

I pick your offer up now.

Thus I have read your research report 8918 (Prequel: A Patch-Like Query Language
for Commit History Search) to some degree.

1. I imagine that the activities behind the description for a chapter like
"5 Compilation of PQL into SmPL" triggered some development challenges.

   Does an interface exists to view the generated SmPL source code?

2. Did this new query language influence further improvements for the semantic
   patch language?

   * Did the error reporting become a bit safer by this command usage?

   * How much will the data format flexibility evolve together with the handling
     of a key word like "flexible" for metavariables?

3. I get concerns from the section "5.2 Idealized syntax".
   How would you categorise it according to the Chomsky-Sch?tzenberger hierarchy?

4. How many means do you offer to restrict commit ranges according to the section
   "6.1 Commit selection"?

5. Can the combination of words like "token" and "regular expression"
   be occasionally confusing?

6. How interesting will discussions become around the used hardware
   that is mentioned in the chapter "7 Evaluation" and the corresponding
   software performance?

7. How is the status for function and class libraries for this application domain?

Regards,
Markus

[Cocci] Querying Git commits with Prequel

[Julia Lawall]

On Thu, 1 Sep 2016, SF Markus Elfring wrote:

> > Prequel is like Coccinelle, but for searching in Git commit histories.
>
> I am curious on how this software approach will evolve further.
>
>
> > http://prequel-pql.gforge.inria.fr/
> >
> > Feedback is welcome.
>
> I pick your offer up now.
>
> Thus I have read your research report 8918 (Prequel: A Patch-Like Query Language
> for Commit History Search) to some degree.
>
> 1. I imagine that the activities behind the description for a chapter like
> "5 Compilation of PQL into SmPL" triggered some development challenges.
>
>    Does an interface exists to view the generated SmPL source code?

It goes into the directory specified with the --tmp argument.


>
> 2. Did this new query language influence further improvements for the semantic
>    patch language?
>
>    * Did the error reporting become a bit safer by this command usage?

No.  The code bases are separate,

>    * How much will the data format flexibility evolve together with the handling
>      of a key word like "flexible" for metavariables?

No idea what this means (data format flexibility).

> 3. I get concerns from the section "5.2 Idealized syntax".
>    How would you categorise it according to the Chomsky-Sch?tzenberger hierarchy?

No idea.  It seems like a straightforward BNF.

> 4. How many means do you offer to restrict commit ranges according to the section
>    "6.1 Commit selection"?

The commit range is just X..Y.  X ad Y could be commit ids or tags (v4.0,
for example).

> 5. Can the combination of words like "token" and "regular expression"
>    be occasionally confusing?

Does the word regular expression occur in the paper?

> 6. How interesting will discussions become around the used hardware
>    that is mentioned in the chapter "7 Evaluation" and the corresponding
>    software performance?

Not sure what there is to discuss.

> 7. How is the status for function and class libraries for this application domain?

No idea what this means.

julia

[Cocci] Querying Git commits with Prequel

[SF Markus Elfring]

>>    * How much will the data format flexibility evolve together with the handling
>>      of a key word like "flexible" for metavariables?
> 
> No idea what this means (data format flexibility).

Is this key word the only one which was added to "PQL" (while it is processed by
"SmPL" in other ways)?


>> 3. I get concerns from the section "5.2 Idealized syntax".
>>    How would you categorise it according to the Chomsky-Sch?tzenberger hierarchy?
> 
> No idea.  It seems like a straightforward BNF.

Does the mentioned syntax belong to the category "context-free"?


> The commit range is just X..Y.  X ad Y could be commit ids or tags (v4.0,
> for example).

Can this range be controlled or adjusted from within Prequel scripts anyhow?


>> 5. Can the combination of words like "token" and "regular expression"
>>    be occasionally confusing?
> 
> Does the word regular expression occur in the paper?

Yes, of course.   ;-)

Is the section "6.1 Commit selection" worth for a few more clarifications?


>> 6. How interesting will discussions become around the used hardware
>>    that is mentioned in the chapter "7 Evaluation" and the corresponding
>>    software performance?
> 
> Not sure what there is to discuss.

Examples:
* Number of processor cores

* Run time distribution for the involved scripts


>> 7. How is the status for function and class libraries for this application domain?
> 
> No idea what this means.

Would you like to share any more information around the evolving code design structure?
Which class libraries will eventually combine software modules?


How do you think about to use another public repository interface
which would be similar to the one for Coccinelle?

Regards,
Markus

[Cocci] Querying Git commits with Prequel

[Julia Lawall]

On Thu, 1 Sep 2016, SF Markus Elfring wrote:

> >>    * How much will the data format flexibility evolve together with the handling
> >>      of a key word like "flexible" for metavariables?
> >
> > No idea what this means (data format flexibility).
>
> Is this key word the only one which was added to "PQL" (while it is processed by
> "SmPL" in other ways)?

There is an exhaustive list of changes in the paper.  To my recollection,
flexible is the only added keyword.  In SmPL, it would cause a parse error
in the place where it appears in PQL.

>
>
> >> 3. I get concerns from the section "5.2 Idealized syntax".
> >>    How would you categorise it according to the Chomsky-Sch?tzenberger hierarchy?
> >
> > No idea.  It seems like a straightforward BNF.
>
> Does the mentioned syntax belong to the category "context-free"?

I guess, to the extent that any programming language grammar is context
free.  Why does it matter?

>
> > The commit range is just X..Y.  X ad Y could be commit ids or tags (v4.0,
> > for example).
>
> Can this range be controlled or adjusted from within Prequel scripts anyhow?

No, it is specified on the command line.

>
> >> 5. Can the combination of words like "token" and "regular expression"
> >>    be occasionally confusing?
> >
> > Does the word regular expression occur in the paper?
>
> Yes, of course.   ;-)

It is used for git.  I'm missing what the issue is.

> Is the section "6.1 Commit selection" worth for a few more clarifications?
>
>
> >> 6. How interesting will discussions become around the used hardware
> >>    that is mentioned in the chapter "7 Evaluation" and the corresponding
> >>    software performance?
> >
> > Not sure what there is to discuss.
>
> Examples:
> * Number of processor cores

1.  As is specified in the paper.

>
> * Run time distribution for the involved scripts
>
> >> 7. How is the status for function and class libraries for this application domain?
> >
> > No idea what this means.
>
> Would you like to share any more information around the evolving code design structure?
> Which class libraries will eventually combine software modules?

I don't even know what a class library is in this context.

>
>
> How do you think about to use another public repository interface
> which would be similar to the one for Coccinelle?

I tried github, but the indices don't fit due to github file size
limitations.  Since I ignore all of the features of github other than the
basic git features, I don't care.  I believe that gforge has a bug
tracker, but I would have to figure out how to enable it.

julia

[Cocci] Querying Git commits with Prequel

[Julia Lawall]

> * Run time distribution for the involved scripts

There is an updated but not currently public version of the paper with this
information.

But in general, the run time depends on the complexity of the script, so
numbers disassociated from what is being expressed are pretty meaningless.

julia

[Cocci] Querying Git commits with Prequel

[SF Markus Elfring]

> I guess, to the extent that any programming language grammar is context free.
> Why does it matter?

Did your research interest grow also for development challenges around
programming languages like "C++" (and their usage in commits)?


>>>> 6. How interesting will discussions become around the used hardware
>>>>    that is mentioned in the chapter "7 Evaluation" and the corresponding
>>>>    software performance?
>>>
>>> Not sure what there is to discuss.
>>
>> Examples:
>> * Number of processor cores
> 
> 1.  As is specified in the paper.

I noticed this fact already.

Will any more software researchers look if they can also afford the mentioned
computation resources so that they will evaluate the shown results a bit more?


> I tried github, but the indices don't fit due to github file size
> limitations.  Since I ignore all of the features of github other than the
> basic git features, I don't care.  I believe that gforge has a bug
> tracker, but I would have to figure out how to enable it.

I am curious if more contributors would appreciate further improvements
around such project management interfaces.

Regards,
Markus

[Cocci] Querying Git commits with Prequel

[SF Markus Elfring]

> It goes into the directory specified with the --tmp argument.

Does the software process "Compilation of PQL into SmPL" generate
any disjunctions (and conjunctions) into the corresponding SmPL source code?

Regards,
Markus

[Cocci] Querying Git commits with Prequel

[SF Markus Elfring]

>> Does the software process "Compilation of PQL into SmPL" generate
>> any disjunctions (and conjunctions) into the corresponding SmPL source code?
> 
> I don't think so.

I am a bit surprised by this feedback.


> In particular, isomorphisms are not applied.

Such special settings might need another look, don't they?


> Why do you ask?

I am just curious on the amount of functionality that is used by the tool "Prequel"
from the semantic patch language.

How much are then also the programming languages "OCaml" and "Python" involved?

Regards,
Markus

[Cocci] Querying Git commits with Prequel

[SF Markus Elfring]

>> How much are then also the programming languages "OCaml" and "Python" involved?
> 
> You can use them as with Coccinelle.  This is mentioned in the paper.

How will their usage evolve together with planned improvements for the tool "Prequel"?

Regards,
Markus